Framing cognitive offloading in terms of gains or losses: achieving a more optimal use of reminders

Lea Fröscher; Ann-Kathrin Friedrich; Max Berentelg; Curtis Widmer; Sam J. Gilbert; Frank Papenmeier

doi:10.1186/s41235-022-00416-3

Framing cognitive offloading in terms of gains or losses: achieving a more optimal use of reminders

Fröscher, Lea; Friedrich, Ann-Kathrin; Berentelg, Max; Widmer, Curtis; Gilbert, Sam J.; Papenmeier, Frank 2022-07-16 00:00:00 Nowadays individuals can readily set reminders to offload intentions onto external resources, such as smartphone alerts, rather than using internal memory. Individuals tend to be biased, setting more reminders than would be optimal. We address the question whether the reminder bias depends on offloading scenarios being framed as either gains or losses, both between‑participants (Experiment 1) and within‑participants (Experiment 2). In both experi‑ ments, framing of reminders in terms of gains resulted in participants employing a risk‑averse strategy and using more reminders than would be optimal. Importantly, however, participants used reminders more optimally and were more willing to choose the risk‑seeking option of remembering internally when reminders implied a loss. Based on metacognitive measures in Experiment 2, the reminder bias increased the more underconfident participants were about their memory abilities in both framing scenarios. Framing did not alter this relationship between erroneous metacognitive underconfidence and reminder bias but provides an additional influence. We conclude that empha‑ sizing the losses (costs) associated with external reminders helps in achieving more optimal decisions in offloading situations, and that in addition to cognitive effort and metacognitive judgments, framing needs to be considered in improving individuals’ offloading behavior. Keywords: Framing, Offloading, Reminders, Metacognition, Prospective memory Public significance statement of the combination of their internal memory resources Faced with the choice of remembering with inter- and the external aids. Therefore, our findings may help to nal memory or using external reminders (e.g., smart- optimize the use of memory aids, that is to compensate phone apps, calendars), individuals generally tend to use for memory limitations in everyday life whenever neces- reminders more than optimal. Our results suggest that sary but at the same time to not neglect internal cognitive emphasizing either the gains (benefits) or losses (costs) resources. associated with external reminders influences individu - als’ decision between using internal memory or exter- Introduction nal reminders. Emphasizing the losses (costs) associated Suppose you have to make an appointment for 4 pm with external reminders led individuals to rely less on tomorrow. Part of everyday life requires remember- external reminders and thus to make more optimal use ing delayed intentions that are fulfilled in the future and stored in prospective memory (Einstein & McDaniel, 1990). However, the capacity of prospective memory is limited (e.g., Cherry & LeCompte, 1999), leading to fre- *Correspondence: frank.papenmeier@uni‑tuebingen.de quent failures in remembering delayed intentions, which Department of Psychology, University of Tübingen, Tübingen, Germany can interfere with functioning in everyday life (Boag Full list of author information is available at the end of the article © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 2 of 18 et al., 2019; Ellis et al., 1999; Kliegel et al., 2000). As tech- intentions maintained over a typical day, including those nology has become commonplace nowadays (e.g., smart- that are trivial or highly practiced such as remembering phones or digital watches), individuals usually have the to go to work, brush one’s teeth, eat, or sleep. An addi- option to offload information onto external devices to tional possible cost of external reminders is that they may help them remember intentions. Using external artifacts prevent an opportunity to strengthen internal cognitive to reduce cognitive demand is known as cognitive off - skills (though see Scarampi & Gilbert, 2020). loading (Risko & Gilbert, 2016). For example, users can readily set up reminders on their smartphones to remind Optimal use of offloading them of their appointment. Considering the decision between relying on offloading Multiple views have been proposed attempting to tools versus internal memory as a tradeoff between costs explain individuals’ choice between relying either on and benefits raises the question of whether there might internal cognitive resources or external aids when solv- be an optimal solution for this tradeoff. With optimality, ing tasks. A minimal memory view proposes that humans we refer to a decision that perfectly balances the costs have the tendency to store information externally when- (e.g., time and effort related to offloading) and the ben - ever possible (Ballard et al., 1995). This is consistent with efits of offloading (e.g., reduced cognitive demand and findings suggesting that individuals aim to avoid actions increased accuracy). Accordingly, we refer to decisions associated with cognitive demand (Kool et al., 2010). as optimal if they maximize the benefits while minimiz - However, this would also imply that individuals should ing the costs of offloading, that is, relying on offloading always use offloading tools regardless of other factors, tools as often as necessary but as seldom as possible. In which is clearly not the case. In contrast, individuals’ turn, we define biases as deviations from this normative choice between either offloading or relying on internal decision-making model. resources is determined by multiple factors, such as To quantify optimality and bias in the context of cogni- memory capacity (Meyerhoff et al., 2021), memory load tive offloading research, Gilbert et al. (2020) introduced (Gilbert, 2015a), monetary reward (Sachdeva & Gilbert, the so-called optimal reminders task. In this task, par- 2020), metacognitive judgments regarding one’s own ticipants are instructed to drag circles to the bottom of internal abilities (Boldt & Gilbert, 2019; Gilbert, 2015b; a square box in ascending numerical order. Sometimes, Risko & Gilbert, 2016; but see Grinschgl et al., 2021a), special circles that were briefly filled with a different color or the interaction and interface design of offloading appeared, and participants had to remember the delayed tools (Grinschgl et al., 2020). Other influence factors on intention of dragging those circles to the correspond- offloading are more generic, such as context (e.g., time ingly-colored border when it was their turn. In order for frame, device) or personal preferences (e.g., personality, participants to fulfill these intentions, they introduced consequences of a missed appointment). them with two strategies: relying on internal memory or The decision to engage in offloading behavior is also setting external reminders. Whereas in some trials, par- affected by cost–benefit considerations (Gray et al., ticipants were forced to use either internal memory or 2006). The most obvious benefit of using offloading tools external reminders, other trials gave participants a free rather than relying on internal resources is that with off - choice between scoring a maximum amount of points loading remembering the offloaded information is nearly using memory or a lesser amount of points using remind- guaranteed (Risko & Gilbert, 2016). This accuracy-related ers. The number of points gained when using reminders benefit of offloading accounts not only for daily life, such was manipulated across trials. Based on participants’ as remembering the items to buy in the grocery with performance in the forced trials, Gilbert et al. (2020) cal- either a shopping list or internal memory. It also accounts culated a normative optimal points value at which par- for well-established offloading tasks, such as the inten - ticipants should switch from using reminders to using tion offloading task (e.g., Gilbert et al., 2020) or the task memory. Based on actual choice behavior, they evalu- of remembering information with the support of writing ated participants’ bias. They observed that participants it down (e.g., Risko & Dunn, 2015). In addition, outsourc- did not show optimal choices. Instead, participants chose ing cognitive demand onto external tools can sometimes reminders more than optimal, thus demonstrating a bias be considered less effortful compared to remembering toward reminders––the reminder bias (Gilbert et al., internally (Ballard et al., 1995; Sachdeva & Gilbert, 2020). 2020). External reminders also incur costs, however. In eve- The size of this reminder bias for delayed intentions ryday life, these costs include the time and effort of set - can be influenced by individuals’ metacognitive judg - ting them up, and the interruptions they can cause. These ments regarding the subjective perception of their costs may be individually minimal. But they would mount internal abilities (Boldt & Gilbert, 2019; Risko & Gil- to an unacceptable level when applied to the multitude of bert, 2016). Reminder use was predicted by individuals’ F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 3 of 18 erroneous underconfidence in their memory abilities outcome (win money vs. loose money), but it occurs also (Gilbert, 2015a; Gilbert et al., 2020). Specifically, the if the same outcome is phrased to appear to involve either reminder bias was higher among those individuals who gains or losses, the so-called framing effect (e.g., Bless were underconfident about their own memory (Enge - et al., 1998; Fagley, 1993; Highhouse & Paese, 1996; for ler & Gilbert, 2020). However, when Gilbert et al. (2020) a review, see Kühberger, 1998; Piñon & Gambara, 2005; corrected participants’ confidence via positive feedback, Steiger & Kühberger, 2018; Tversky & Kahneman, 1981). thus aiming at debiasing participant’s offloading choices For example, under a gain framing, individuals would by making them less underconfident, participants were prefer the risk-averse option of winning $300 with cer- less biased toward using reminders but still offloaded tainty over the risky option of winning either $1000 with more than would be optimal. In a similar vein, provid- 30% probability or $0 with 70% probability. If the same ing participants with a financial incentive based on over - outcome (average profit of $300) is framed under losses, all performance reduced the reminder bias, thus causing individuals show a shift to a risk-seeking strategy. That is, participants to use reminders more optimally (Sachdeva if one tells individuals that they have an initial $1000 and & Gilbert, 2020), likely because participants were will- that they could choose between the option of losing $700 ing to invest more cognitive effort. However, despite the with certainty and the option of losing either $0 with 30% reduction in the reminder bias, this was again not enough probability or $1000 with 70% probability, individuals to debias participant’s offloading choices. Thus, if used as would prefer the later (risky) option. a tool to achieve a more optimal use of reminders, both Applying those findings from the literature on deci - approaches were only partially effective in reducing the sion-making (Kahneman & Tversky, 1979) to cognitive reminder bias, suggesting that there must be further offloading makes evident that the previously reported barriers that first have to be overcome in order for the reminder bias (Gilbert et al., 2020) could also be seen reminder bias to be fully eliminated. as individuals employing a risk-averse choice strategy. Within previous research on the reminder bias (Engeler Offloading vs. internal memory: decision & Gilbert, 2020; Gilbert et al., 2020; Sachdeva & Gilbert, under uncertainty 2020), the choice between using reminders or internal We take a novel perspective on individuals’ offloading memory was framed in terms of gains, such as earning decisions. Considering the typically achieved perfor- some points using reminders and earning a maximum mance when using an offloading strategy compared to reward using memory. Applying the certainty effect to internal memory, it turns out that performance is usually cognitive offloading, one would thus expect that individ - very high or even near perfect with offloading compared uals prefer outcomes that are near-certain (using an off - to a much more unreliable performance when relying on loading strategy) over outcomes that are more uncertain the internal memory. u Th s, when choosing between off - (using internal memory), thus resembling the reminder loading or internal memory, individuals decide between bias. achieving a relatively certain outcome when using the off - Given the idea that the choice between cognitive off - loading strategy and an uncertain or risky outcome when loading and internal memory represents a decision under using internal memory. uncertainty, a switch from gain framing to loss framing When making decisions under risk, individuals typi- should also cause a switch from risk-averse choice strate- cally prefer certain over uncertain choices, also known gies to risk-seeking choice strategies within the context as the certainty effect (Kahneman & Tversky, 1979). For of cognitive offloading. Thus, with the aim of using the example, individuals would prefer winning $3000 with framing of reminders as a means of achieving a more certainty over winning $4000 with 80% probability, even optimal use of reminders, reversing the framing from though the expected outcome is higher in the second gains to losses might reduce, eliminate, or even reverse variant. Interestingly, this bias reverses when the choice the reminder bias. With the present research, we inves- options are presented as losses rather than gains, also tigated whether this shift in the reminder bias could be known as the reflection effect (Kahneman & Tversky, observed. 1979). For example, individuals would prefer losing $4000 with an 80% probability over losing $3000 with certainty, Experiment 1 even though the expected loss is lower in the second vari- Our first experiment used the optimal reminders task ant. That is, switching from gains to losses also causes a (Gilbert et al., 2020) and expanded it by implementing a shift from risk-averse choice strategies to risk-seeking between-participants manipulation regarding the fram- choice strategies (e.g., Baucells & Villaís, 2010; Kahneman ing of reminders. We framed reminders as gains, just as & Tversky, 1979; Mather et al., 2012). This shift in choice previous research, or as losses, which had not been done strategies occurs not only if there is an actual change in before. We had two key hypotheses. In the gain framing Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 4 of 18 condition, we expected to replicate previous findings on cognitive offloading, thus showing a reminder bias (Gil - bert et al., 2020). For the loss condition, we expected a reduction of the reminder bias. This experiment was pre - registered; see https:// osf. io/ rcu8v. Method Participants We recruited the participants using the student mailing list of the University of Tübingen and online posts shared in non-commercial Facebook and WhatsApp groups. Following exclusions (see below), our sample consisted of 141 participants (94 reporting their gender as male, 32 as female, and 2 as other). Participants had a mean age of 26.30 years (SD = 8.90, range = 18–64). The experi - ment took approximately 45 min, for which participants received the opportunity to win one of six €10 gift cards from a local bookstore or train company. We conducted Fig. 1 Illustration of the optimal reminders task. Note Example trial of the optimal reminders task. a Participants were instructed to drag this experiment in accordance with the APA guidelines circles to the bottom edge of a box in sequential order. Each time a for research ethics, and participants provided informed circle was moved onto an edge, it disappeared from the screen and consent before participating. the next circle in sequence emerged; b Sometimes, new circles were We performed a power analysis based on the results initially highlighted in a different color, indicating a delayed intention observed in the unadvised group of Experiment 2 in Gil- to drag the special circle to the same‑ colored edge when reached in sequence; c A special circle’s color faded back to yellow two seconds bert et al. (2020), using the R package powerbydesign after appearance. d If permitted, participants set reminders by (Papenmeier, 2018). To achieve a power of 80% for the instantly dragging the special circles near their intended edge when investigated interaction effect (assumption of reminder they emerged on the screen; e Participants carried on with the task of bias under gain framing and no bias under loss framing; dragging circles to the bottom of the box; f After dragging the circles see the script containing the power simulation for details: in sequential order, they could then execute the delayed intention to drag the special circle to its intended location https:// osf. io/ xqt8j), we required a sample size of 136 participants. We stopped data collection after two weeks, with the study slightly overshooting the targeted sample size (N = 141, 9 excluded) at this point in time. toward using or their own memory, or (c) optimally calibrated. On each trial, participants used their computer mouse Optimal reminders task to sequentially drag 25 numbered circles to the bottom We modified the optimal reminders task used by Gilbert of a box (Fig. 1). Up to six circles were visible a time, et al. (2020). In this task, participants can choose (a) to and each time a circle was removed from the box, it was remember intentions using internal memory, which replaced with a new one (e.g., after dragging ‘1’ to the leads to a maximum reward for each remembered item, bottom, a new circle labeled ‘7’ appeared in its place). The or (b) they can set external reminders, which leads to a left, top, and right edges of the box were colored blue, smaller reward that varies from trial to trial. This para - orange, and purple, respectively. Occasionally, new cir- digm allowed us to examine not only the frequency of cles appeared initially in one of these colors before fad- reminder-setting but also its optimality. For example, ing to yellow after 2 s. This was an instruction to form suppose an individual’s accuracy is 55% when using their a delayed intention to drag these ‘special’ circles to the own memory and 100% when using reminders. If they are corresponding edge of the box. For example, if a spe- given a choice between earning 10 points per item using cial circle (e.g., 7) initially appeared as blue, participants their own memory or 5 points per item using reminders, needed to remember this instruction while they dragged it is optimal to use the internal memory strategy. But if circles 2 to 6 to the bottom of the box (by which time they are offered 6 points per item when using reminders, the special circle had faded to yellow). They could then it is optimal to select this strategy instead. By compar- execute the intention to drag 7 to the left. Within each ing participants’ choices with the optimal strategy, this trial consisting of 25 circles, 10 special circles were pre- paradigm can be used to calculate whether individuals sented. These circles appeared between the 7th and 25th are (a) biased toward using external reminders, (b) biased F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 5 of 18 circle in the sequence. (The initial 6 circles were already the width/height of the box, and all circles were initially on screen at the beginning of the trial, so they could not placed so that they fall within a central portion of the box act as targets.) These 19 possible target positions were with dimensions sized at 56% of the total width/height, split into 10 adjacent bins (9 of which had a length of two so that no circles were adjacent to any of the edges of the and one of which, placed randomly in the sequence, had box at the beginning of the trial. length one). One target was then placed randomly within each of these bins. As a result of the multiple concur- Procedure rent intentions, participants were unlikely to remember After providing informed consent, the computer ran- all if they relied on internal memory alone. Alternatively, domly assigned participants to one framing condition, if they used reminders, they could offload the inten - with the assignment resulting in 63 and 78 participants tions by immediately dragging special circles next to for the gain and loss framing, respectively. Participants the instructed edge when they first appeared (e.g., drag - then completed the six practice trials: In the first trial, ging a blue 7 toward the left edge of the box as soon as participants dragged the circles in sequence to the bot- it appeared rather than waiting for it to fade to yellow tom of the square (8 circles in total). They further prac - first). The location of the special circle then acted as a ticed by dragging one special circle to the instructed reminder when the participant reached this number in edge in the second practice trial (8 circles in total). They the sequence. An everyday analogy would be leaving an repeated this practice trial and were not allowed to con- object by the front door so that you remember it when tinue the experiment until they responded correctly to leaving the house the next day. this special circle. They then continued with two con - Participants alternated between ‘forced’ and ‘choice’ secutive practice trials of actual length (10 special circles trials. On forced trials, they had to use either their own out of 25 circles in total). Following this, they were made memory (‘forced-internal’) or reminders (‘forced-exter- aware of the ability to use reminders in this task. They nal’). On choice trials, participants decided between practiced again, but on the forced-external trial type. This earning 10 points per remembered item (using their own time, they needed to respond correctly to at least 8 out of memory), or a smaller number of points between 1 and 10 special circles in order to continue the experiment. In 9 (using external reminders). We calculated the optimal line with previous studies (Gilbert, 2015b; Gilbert, et al., strategy based on performance on the forced trials, then 2020), we used the exact value of eight correct responses compared this with their actual decisions on the choice to ensure that participants were able to achieve at least trials. The experiment was split into two conditions (gain 80% accuracy with using the external strategy. After per- and loss), with participants being assigned randomly to forming one additional forced-external practice trial, one condition. In both conditions, participants were participants were instructed about the upcoming forced given 0 points at the beginning of the experiment. Fur- and choice trials and that their task was to gain as many ther, participants in the gain condition chose between points as possible (lose as few points as possible). receiving 10 points for each remembered special circle or During the main experiment, participants performed a smaller number of points (1–9) to use reminders (see a total of 17 trials. On odd-numbered trials, participants Fig. 2). For each missed special circle, participants gained were given a free choice between using internal memory zero points. This matches the version of the task used in (gain 10 points/lose 0 points; according to framing con- previous research (Gilbert et al., 2020). In the loss condi- dition) or reminders (gain/lose 1–9 points per special tion, participants lost 10 points for each missed special circle, presented in random order). On even-numbered circle, and they had the choice between (a) using their trials, participants alternated between the forced-exter- own memory and keeping all their previously scored nal and forced-internal trials, with the starting trial type points (losing 0) each time they correctly remembered (external or internal) randomized between participants. special circles or (b) using reminders and scoring minus The trial number was set to the exact number of 17 (9 points every time they remembered (− 9 to − 1). All free, 8 forced) trials to be consistent with previous ver- instructions were presented in German. sions of the optimal reminders task (Gilbert et al., 2020). After completion, participants were given the opportu- Apparatus nity to enter the prize draw on SoSci Survey. Participants completed the task via their computer’s web browser. Participation was only permitted if the browser Reward window had dimensions of at least 500 × 500 pixels. The Participants were told that they were scoring points, square box containing the circles was sized at 80% of with the prospect of earning (losing) up to 1700 points the horizontal or vertical extent of the browser window, in the gain (loss) condition. Therefore, the earnings whichever was smaller. Each circle had a radius of 5.5% of could range between 0 and 1700 points in the gain Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 6 of 18 Fig. 2 Example instructions for the free‑ choice trials. Note Prior to the start of each free‑ choice trial, participants were given the choice to either rely on their internal memory or to set external reminders. a In the gain condition, participants would score 10 points for each special circle they remembered if they relied on their memory or a smaller number of points (1–9) if they selected to set reminders. b In the loss condition, if participants relied on their memory, they would lose 0 points for each special circle they remembered. If they chose to use reminders, however, they would score minus points (− 9 to − 1). For both framing conditions, a sequence of 25 circles was presented in each trial, with 10 of them acting as special circles condition, and between minus 1700 points and 0 points • Forced-internal accuracy (ACC ). This is the mean FI in the loss condition. The experiment was promoted by target accuracy (proportion of special circles cor- offering participants the chance to win one of six 10€ rectly dragged to the instructed location) on forced- gift cards for taking part in the experiment. internal trials. • Forced-external accuracy (ACC ). This is the mean FE target accuracy (proportion of special circles cor- rectly dragged to the instructed location) on forced- Design external trials. The experiment employed a 2 (framing condition: gain × • Optimal indifference point (OIP). For choice tri - vs. loss) 2 (indifference point: optimal vs. actual) als, this is the value for special circles offered with design with framing as a between-participants variable, reminders at which an unbiased individual should and we defined five variables of interest: F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 7 of 18 be indifferent between the two options, based on the it earned them fewer points, indicating random strat- accuracy in the forced-internal and forced-external egy selection; n = 8); and (e) their reminder bias score trials (ACC and ACC ). As in Gilbert et al. (2020), exceeded 2.5 standard deviations from the group mean FI FE this was calculated as (n = 1). The data are publicly available (https:// osf. io/ 8shkf/). OIP × ACC = 10 × ACC (1) FE FI Transparency and openness If the OIP was less than 1 or greater than 9, it was For the study’s entire research report, we communi- set to the relevant lower or upper bound. This was cate our methodical and statistical approach including so that the potential values of the OIP would match sample size estimation, data exclusion, experimental the potential values of the point at which they were manipulation, and measures of interest. All hypotheses, actually indifferent, which was bound by their experimental methods, and planned analyses were pre- choices for values 1 to 9. registered before data collection. All analyses were run in R, and information on the R environment (including package versions) used for the analyses is given in the • Actual indifference point (AIP). This is the esti - analysis script. mated point for choice trials at which participants were actually indifferent to the two strategy options. As in Gilbert et al. (2020), this was calculated by fit - Results ting a sigmoid curve to the strategy choices (0 = own Accuracy memory; 1 = reminders) across the nine special val- Participants were able to remember almost two-thirds of ues (1–9), using the quickpsy function from the R the special circles using their memory in forced-internal package quickpsy (Linares & López-Moliner, 2019) trials, but nearly all of them using reminders in forced- bounded to the range 1 to 9 (see the analysis script external trials (see Fig. 3a). This data is consistent to on OSF; https:// osf. io/ qsfmy/). Based on this curve, the accuracy data found in Gilbert et al. (2020). As an we were able to estimate the point associated with exploratory analysis, we submitted the accuracy data to 50% probability of choosing either strategy, which is a 2 (condition: forced-internal vs. forced-external; within) the AIP. × 2 (framing: gain vs. loss; between) mixed analysis of • Reminder bias. This is defined as OIP–AIP, which variance (ANOVA). Whereas the main effect of condition will yield a positive value for a participant biased was significant, F(1, 139) = 641.22, p < 0.001, η = 0.82, toward using more reminders than would be optimal, neither the main effect of framing, F(1, 139) = 0.03, and a negative value for a participant biased toward p = 0.856, η < 0.01, nor the interaction of framing using fewer reminders than would be optimal. and condition, F(1, 139) = 0.02, p = 0.899, η < 0.01, was significant. Thus, although participants showed a Each of the previous five measures was calculated sepa - higher task accuracy when using reminders, their over- rately for the gain and loss condition, and to compare all task performance was not affected by the framing of AIP and OIP between framing conditions, calculation of reminders. indifference points was performed after transforming the minus points from the loss framing condition by adding Reminder bias As defined in our preregistration, we investigated the influence of framing on the reminder bias (defined as OIP Exclusion criteria minus AIP) by submitting the reminder bias scores to a In accordance with our preregistration, we excluded 2 (framing condition: gain vs. loss; between) × 2 (indif- participants if (a) their accuracy in forced-internal tri- ference point: OIP vs. AIP; within) mixed ANOVA (see als (averaged across gain and loss conditions) was lower Fig. 3a). There was a significant main effect of indif - than 10% (n = 0); (b) accuracy in the forced-external tri- ference point, F(1, 139) = 52.17, p < 0.001, η = 0.27, als was lower than 70% (averaged across gain and loss and a non-significant main effect of framing condition, conditions; n = 0); (c) accuracy in the forced-internal tri- F(1, 139) = 3.45, p = 0.065, η = 0.02. Importantly, the als was higher than in the forced-external trials in either interaction of framing and indifference point was sig - condition (n = 0); (d) there was a negative point biserial nificant, F(1, 139) = 10.05, p = 0.002, η = 0.07, with the correlation between points offered for correct responses reminder bias scores, on average, 1.19 points greater in on each trial using reminders (1–9) and choice of strat- the gain (M = 2.00, SD = 2.45) compared to loss condi- egy (0 = own memory, 1 = reminders; this excluded par- tion (M = 0.81, SD = 2.00). We further analyzed whether ticipants who were more likely to set reminders when the reminder bias scores were significantly greater than Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 8 of 18 Fig. 3 Results on Offloading Behavior from Experiment 1. Note a In the left panel, mean accuracy for the forced‑internal and forced‑ external trials in the gain and loss framing conditions. Mean optimal (OIPs) and actual indifference points (AIPs) as a function of framing condition in the right panel. The line pattern illustrates the size of the reminder bias defined as OIP minus AIP. b Correlation between actual and optimal indifference points for the gain (left panel) and loss (right panel) framing condition. Error bars represent standard errors zero with preregistered one-tailed paired t tests (OIPs Fisher’s (1925) z = − 2.53, p = 0.011. That is, despite par - vs AIPs). Although framing reminders in terms of losses ticipants having a systematic bias toward using remind- instead of gains resulted in a significant reduction of the ers, those who derived the most benefit from reminders reminder bias, the reminder bias was still significant both (i.e., low OIP) were also most likely to use them (i.e., in the gain, t(62) = 6.49, p < 0.001, d = 0.82, and loss con- low AIP) and, although evident in both conditions, this dition, t(77) = 3.60, p < 0.001, d = 0.41. That is, although relationship was more pronounced when offloading was participants offloaded more optimally under loss framing associated with a loss (see Fig. 3). than under gain framing, participants still offloaded more than optimal also under loss framing. Discussion We ran an exploratory analysis on the relationship Consistent with our hypotheses, we observed a strong between the OIPs and AIPs. To achieve this, we calculated reminder bias in the gain condition of Experiment 1. That a Pearson’s product–moment correlation on the indiffer - is, participants offloaded more than optimal, replicat - ence points for each framing condition. We further tested ing previous findings (Gilbert et al., 2020). In addition, for the difference between correlations by using the R we showed for the first time that the framing of remind - package cocor (test of significance for independent cor - ers in terms of gains or losses shapes the reminder bias. relations; Diedenhofen & Musch, 2015). There was a sig - That is, the reminder bias was largely reduced under loss nificant positive correlation between the OIPs and AIPs framing. This is in line with the literature on decision- in both conditions (gain: r = 0.29, p = 0.019; loss: r = 0.63, making (Kahneman & Tversky, 1979), as it suggests that p < 0.001; see Fig. 3b), with the correlation being signifi - while individuals employ a risk-averse strategy in situ- cantly higher in the loss compared to the gain condition, ations involving gains, they are more inclined to take a F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 9 of 18 risk in situations involving losses. However, it is to note studies in this meta-analysis, at least 49 participants were that despite participants being more risk-seeking when required for 80% power ( α = 0.05, two-tailed paired t facing the prospect of a loss, they still deviated from opti- test). This experiment also provided data for a separate mal offloading behavior in the loss framing condition, unrelated project, which was reported elsewhere (Kirk thus using reminders more than optimal. This finding is et al., 2021). Therefore, we were aiming for a higher sam - consistent with the view that also other factors influence ple size (i.e., N = 300), as this was the intended sample the reminder bias, such as participants’ underconfidence size for the other project. As in earlier studies (Gilbert, in their memory abilities (Gilbert et al., 2020) and their 2015a, 2015b), participation was restricted to volun- invested cognitive effort (Sachdeva & Gilbert, 2020). We teers aged at least 18 years, located in the USA. We also addressed these points in Experiment 2. restricted inclusion to participants with a minimum of 90% Mechanical Turk approval rate. Participation took Experiment 2 approximately 45 min, for which participants were guar- In Experiment 2, we again manipulated the framing of anteed a base payment of $2, plus an additional bonus reminders as gains or losses, while participants per- of up to $8.67 depending on their task performance. formed the optimal reminders task. In addition, we asked Participants had a mean age of 37.81 years (SD = 10.97, the participants to make metacognitive judgments at the range = 21–72); 190 reported their gender as male, 108 beginning of the experiment. This allowed us to assess as female, and 2 as other. Ethical approval was received their overconfidence or underconfidence in carrying from the UCL Research Ethics Committee (1584/003) out the task (Gilbert et al., 2020). Further, we rewarded and participants provided informed consent before par- participants depending on their task performance. This ticipating in the study. acted as a financial incentive to increase both the cog - nitive effort that participants invest while performing Optimal reminders task the task and the optimality of their offloading choices We used the same modified version of the optimal (Sachdeva & Gilbert, 2020). Furthermore, we manipu- reminders task as in Experiment 1, with small adapta- lated framing within-participants instead of between- tions. First, Experiment 2 used an English-speaking participants in order to account for potential individual MTurk sample, and thus, all items were presented in differences in framing effects (see Levin et al., 2002). As English. Second, we included metacognitive accuracy in Experiment 1, we expected to observe a reminder bias judgments in between the practice trials. That is, both in the gain framing condition, but a reduction (or even after practicing trials on the forced-internal and the reversal) of this bias in the condition where reminders forced-external type, participants provided a measure implied a loss. This experiment was preregistered; see of how confident they were at their ability to perform https:// osf. io/ 8zvf6/. the task (see Fig. 4). Third, framing was implemented as a within-participants variable. This allowed us to test Method whether results observed in Experiment 1 can also be Participants observed when comparisons between framing conditions Participants were recruited from the Amazon Mechani- are made within, rather than between individuals. To cal Turk website (http:// www. mturk. com), an online do this, the task was split into two blocks: gain and loss marketplace in which participants receive payment for (with the order of blocks being counterbalanced between completion of web-based tasks (Crump et al., 2013). participants). Sample size was estimated performing a power analysis Finally, we did not assign negative values to loss points. with G*Power 3.1 (Faul et al., 2007). The power analysis That is, while instructions of the gain condition stayed was based on the meta-analysis of Kühberger (1998), as the same, during the loss condition participants received to our knowledge there was no previous study that has the maximum of points available for this block (100 investigated the effect of gain and loss framing within- points per trial) before the beginning of the block. They participants in the context of cognitive offloading. With were presented with the choice between either using a Cohen’s d effect size of 0.41 for the within-participants their own memory and keeping all their points (losing 0) each time they correctly remembered special circles, or using reminders, and losing points every time they At this point, we want to clarify that both experiments were conducted remembered (2–8). Contrasting to Experiment 1, this independently from two separate working groups. We follow the same guide- had the advantage that both conditions were equiva- lines of transparency and openness as in Experiment 1; however, some parts lent in terms of the reward participants received, that is of the methodical and statistical approach differ between experiments. Also, note that in the original order Experiment 2 has been performed prior to the outcomes were phrased to appear as either gains or Experiment 1. We changed experiments within the scope of the current paper losses, but objectively they were the same. For example, if as we perceived this order to be more coherent. Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 10 of 18 Fig. 4 Metacognitive Confidence Judgment. Note The metacognitive confidence rating provided us with our metacognitive confidence measure: After participants finished a series of practice trials, they rated their confidence in their ability to accurately remember the delayed intention of dragging the special circles to the respective edges. Participants performed these ratings once (prior to the start of the experiment) separately for the internal and external strategies. For the confidence rating after the forced‑ external trials, participants received the alternate instruction ’Now that you have practiced doing the task using reminders, we would like you to tell us how accurately you can perform the task when you use this strategy’ offered 7 points to use reminders and a participant chose circle, presented in random order). On even-numbered to use reminders and successfully remembered every trials, participants alternated between the forced- special circle, in the gain condition they would earn 70 external and forced-internal trials, with the starting points. Whereas, in an equivalent trial of the loss con- trial type (external or internal) randomized between dition (i.e., 3 points offered to use reminders), they lost participants and counterbalanced between gain/loss 30 points and thus also retained 70 points. By using this conditions. Participants then received experimental scoring scheme, we were expecting to maximize the com- instructions for the other condition (gain or loss). After parability of framing conditions, thus reducing the risk finishing the second experimental block (gain or loss; of potential data noise. Note that this change in scoring 13 trials as above), participants completed two ques- also resulted in Experiment 2 studying the framing effect tionnaires. This was part of the other unrelated project, whereas Experiment 1 investigated the reflection effect addressing a different question which is reported in a (Fagley, 1993). separate paper (see Kirk et al., 2021, for full details). For a demonstration, the entire experiment can be accessed at. Procedure http:// ucl. ac. uk/ sam- gilbe rt/ demos/ CWPK1/ start. html Participants first provided informed consent and then completed six practice trials, with the practice trials following the same procedure as in Experiment 1. How- Reward ever, after each pair of consecutive practice trials with This time we paid participants depending on their task and without reminders, participants made their meta- performance. Paying participants based on their task cognitive judgments reporting how accurately they can performance should ensure that they are more moti- perform the task with the respective strategy. Partici- vated to make optimal choices (see Sachdeva & Gilbert, pants were then randomly assigned to the gain or loss 2020). Implementing this payment allowed us to evaluate condition. In the first experimental block (gain or loss), whether effects of framing observed in Experiment 1 dif - participants performed a total of 13 (7 free, 6 forced) fer in situations where participants have a financial incen - trials. Due to the within-participants implementation tive to choose optimally. To do this, participants were in Experiment 2, participants had to perform trials of told that they were scoring points, where 300 points were both conditions and therefore for practical reasons, equivalent to $1. They received 600 points at the begin - that is to keep the task duration within reasonable ning of the experiment. Then, they were additionally able limits for the participants, we reduced the total trial to earn (or keep) up to 1300 points (i.e., 100 points per number per block. As a result of that, the range of the trial) in each half of the experimental trials. Therefore, indifference points were changed in accordance with the earnings could range between 600 points ($2) and the reduced number of the free-choice trials, that is 3200 points ($10.67). The experiment was advertised as from 1 to 9, in Experiment 1, to 2 to 8, in Experiment having a base payment of $2, which participants received 2. On odd-numbered trials, participants were given a simply for taking part, with the additional earnings sent free choice between using internal memory (10 points to participants afterward as a bonus payment. per special circle) or reminders (2–8 points per special F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 11 of 18 Design and they remembered nearly all of the special circles We entered framing (gain vs. loss) as within-partici- when using reminders in forced-external trials (see pants variable into our design. The variables of interest Fig. 5a). This replicates the accuracy data found in were the same as in Experiment 1, with the indifference Experiment 1 and previous research (Gilbert et al., 2020). points bounded to the range 2 to 8. In addition, measures We submitted the accuracy data to a 2 (condition: forced- regarding the metacognitive judgments were added: internal vs. forced-external; within) × 2 (framing: gain vs. loss; within) ANOVA with repeated measures on both • Internal metacognitive confidence. This is the variables. We again obtained a significant main effect of response made to the metacognitive accuracy predic- condition, F(1, 299) = 991.36, p < 0.001, η = 0.77, but tion following practice trials using internal memory no significant main effect of framing, F(1, 299) = 0.33, (see Fig. 4). p = 0.565, η < 0.01, and a non-significant interaction • External metacognitive confidence. This is the of framing and condition, F(1, 299) = 0.02, p = 0.892, response made to the metacognitive accuracy predic- η < 0.01. This indicates that accuracy was significantly tion following practice trials using reminders. affected by the use of reminders, but not by the framing • Internal metacognitive bias. This is the difference of reminders, just as in Experiment 1. between metacognitive confidence and actual accu - racy on forced-internal trials. A positive number Reminder bias would indicate overconfidence of their own memory Our key hypotheses were tested using the reminder bias abilities. scores. As preregistered, we analyzed the reminder bias • External metacognitive bias. This is the difference scores (OIP minus AIP) with two-tailed one-sample t between metacognitive confidence and actual accu - tests. We observed a significant reminder bias both in the racy on forced-external trials. A positive number gain framing condition, t(299) = 10.32, p < 0.001, d = 0.60, would indicate overconfidence of their performance and in the loss framing condition, t(299) = 4.37, p < 0.001, when using reminders. d = 0.25. To evaluate whether these reminder bias scores differed as a function of framing conditions, we per - formed a preregistered two-tailed paired t test comparing Exclusion criteria reminder bias scores of the gain and loss conditions. This Similar to Experiment 1, participants were excluded if (a) analysis yielded a reminder bias that was, on average, 0.70 their accuracy in the forced-internal condition was lower points greater in the gain condition (M = 1.28, SD = 2.14) than 10%, averaged across the gain and loss conditions; compared to the loss condition (M = 0.58, SD = 2.28), (b) accuracy in the forced-external condition was lower t(299) = 5.05, p < 0.001, d = 0.29. than 70%, averaged across the gain and loss conditions; Given that we manipulated framing within-participants (c) accuracy on the forced-internal trials was higher than in Experiment 2, we obtained two separate reminder bias forced-external trials in either condition; (d) there was a scores for each participant, once for the gain condition negative point biserial correlation between points offered and once for the loss condition. We calculated a prereg- for correct responses on each trial using reminders (2–8) istered Pearson’s product–moment correlation on these and choice of strategy (0 = own memory, 1 = reminders; two scores to check whether individual differences in task this excludes participants who were more likely to set behavior were significantly related across framing condi - reminders when it earned them fewer points, suggest- tions. We found that the reminder bias scores in the two ing random strategy selection); (e) reminder bias score framing conditions were significantly correlated, r = 0.41, (averaged across the gain and loss conditions) exceeded 3 p < 0.001. This indicates that low (high) reminder bias median absolute deviation units (MAD; Leys et al., 2013); scores in one framing condition were associated with low (f ) difference in reminder bias scores between the two (high) scores in the other condition. As in Experiment 1, conditions exceeded 3 MAD units; and (g) internal meta- we exploratorily performed a Pearson’s product–moment cognitive bias score exceeded 3 MAD units. Data collec- correlation between the OIPs and AIPs for both framing tion continued until the study had the appropriate power conditions (see Fig. 5b). There was a significant correla - (N = 300) following exclusion (64 excluded). The data are tion both under gain framing, r = 0.43, p < 0.001, and loss publicly available (https:// osf. io/ 8zvf6/). framing, r = 0.38, p < 0.001, with the difference between correlations being non-significant, Pearson and Filon’s Results Accuracy Please note that Kirk et al. (2021) used the data of the gain condition for Participants remembered almost two-thirds of the spe- their analyses. Therefore, some values concerning offloading behavior in the cial circles using their memory in forced-internal trials, gain condition are identical between the two projects. Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 12 of 18 Fig. 5 Results on Offloading Behavior from Experiment 2. Note a In the left panel, mean accuracy for the forced‑internal and forced‑ external trials in the gain and loss framing condition. Mean optimal (OIPs) and actual indifference points (AIPs) as a function of framing condition in the right panel. The line pattern demonstrates the size of the reminder bias defined as OIP minus AIP. b Correlation between actual and optimal indifference points for the gain and loss framing condition. Error bars represent standard errors (1898) z = 0.83, p = 0.405. (Note that Experiments 1 and and M = − 11.21, SD = 14.97, respectively). This indi - 2 differ in this respect as this time we tested for the dif - cated that participants were significantly underconfident ference between dependent, non-overlapping correla- about their memory abilities and their accuracy using tions using the R package cocor; Diedenhofen & Musch, reminders. Second, we assessed whether participants’ 2015.) Similar to Experiment 1, participants who ben- underconfidence in their internal judgments was associ - efited the most from using reminders (i.e., low OIP) were ated with a higher reminder bias by calculating a Pear- more likely to offload intentions (i.e., low AIP). However, son’s product–moment correlation between the internal in contrast to Experiment 1, this relationship between metacognitive bias score and the reminder bias score actual task performance and offloading behavior did not for both framing conditions (see Fig. 6). A significant differ between framing conditions. Metacognitive bias Note that according to our preregistration we planned to correlate the inter- nal metacognitive bias score with the reminder bias within each condition As reported in the preregistration, we performed three (gain or loss). However, we realized that this could yield spurious self-correla- additional analyses involving metacognitive judgments. tions due to non-independent observations, because the forced-internal accu- First, we performed separate two-tailed one-sample t racy score was used to calculate both the internal metacognitive bias and the reminder bias. To avoid this potential statistical artifact, we deviated from the tests on the internal and external metacognitive bias preregistration by calculating the internal metacognitive bias from the oppo- scores against zero to test whether participants were site experimental condition (i.e., the reminder bias in the gain condition was under- or overconfident using the two strategies. Both correlated with the internal metacognitive bias from the loss condition, and vice versa). This ensured that the correlation analyses met the assumption of the internal, t(299) = − 3.65, p < 0.001, d = 0.21, and independent observations. Results were similar when the analyses were con- external, t(299) = − 12.96, p < 0.001, d = 0.75, metacogni- ducted as specified in the original preregistration. tive bias scores were significant (M = − 6.69, SD = 31.76; F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 13 of 18 Fig. 6 Relationship between Metacognitive Confidence and Offloading Behavior. Note Correlation between individuals’ metacognitive bias and reminder bias scores, separately for both framing conditions. Negative metacognitive bias scores illustrate underconfidence in memory ability and positive reminder bias scores represent overuse of reminders negative correlation was found in both framing condi- p < 0.001. We thus selected the model with the two fixed tions between reminder bias and internal metacognitive factors and compared it with a model including the inter- judgments (gain: r = − 0.23, p < 0.001; loss: r = − 0.19, action of internal metacognitive bias and framing as the p < 0.001), indicating that underconfident participants third fixed effect (AIC: 2603.60), with no significant dif - were more biased toward reminders, regardless of fram- ferences between the models in a likelihood-ratio test, ing condition in free-choice trials. Third, we assessed χ (1) = 0.09, p = 0.761. u Th s, we retained the simpler whether the relationship between reminder bias and model (additive terms only). This suggests that framing internal metacognitive bias differed as a function of adds an additive predictive value to the reminder bias, framing condition by using the R package cocor (test for that is beyond the effect of participants’ metacognitive dependent, non-overlapping correlations; Diedenhofen miscalibration of their internal memory abilities. & Musch, 2015). We found no evidence that framing altered the relationship between internal confidence and Discussion the tendency to use reminders, Fisher’s (1925) z = − 0.62, Consistent with our study’s key hypotheses, Experi- p = 0.537. ment 2 illustrates that while participants choose to use In an exploratory analysis, we further investigated reminders more than would be optimal, this systematic whether framing influences the reminder bias over bias is reduced by the framing of reminders in terms of and above participants’ internal metacognitive bias. To losses. This indicates that, despite receiving financial achieve this, we fitted mixed-effects models (R package incentive to offload optimally, individuals employ a risk- lme4; Bates et al., 2015) on the reminder bias scores with averse approach in situation involving gains but a more random intercepts for the participant effect. To initiate risk-seeking strategy when offloading implies a loss (see the model selection process, we compared the intercept- Kahneman & Tversky, 1979). Furthermore, in the scope only model (Akaike information criterion [AIC]: 2630.50) of participants’ metacognitive confidence ratings, Experi - with a model with internal metacognitive bias as continu- ment 2 demonstrates that irrespective of the framing of ous fixed factor (AIC: 2623.20). We found significant dif - reminders individuals are underconfident about their ferences between the models in a likelihood-ratio test, χ memory abilities and that this erroneous metacognitive (1) = 9.29, p = 0.002, indicating that the reminder bias underconfidence in turn is associated with a greater bias was significantly increased by participants’ undercon - toward using reminders in both framing scenarios, which fidence in their memory abilities We thus selected the replicates the negative relationship between internal model with the one fixed effect and compared it with a metacognitive bias and reminder bias found in Gilbert model which also included the framing (gain vs. loss) as et al. (2020). Further, we show that the framing of remind- another fixed factor (AIC: 2601.70). Again, a likelihood- ers influences participants choice in using reminders over ratio test revealed significant differences, χ (1) = 23.51, and above participants’ internal metacognitive bias. Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 14 of 18 General discussion (Kahneman & Tversky, 1979), as it demonstrates that The present research aimed at achieving more optimal a switch from gain to loss framing also causes a switch decisions in offloading situations, by assessing whether from a risk-averse strategy to a more risk-seeking strat- the framing of reminders (Kahneman & Tversky, 1979) egy. Moreover, our study illustrates that this switch in influences how optimally intentions are offloaded onto strategy also applies to cognitive offloading. external devices. In two experiments, we manipulated However, even though participants were more risk- the framing of reminders in terms of gains or losses, once seeking in situations involving losses, in both experi- between-participants (Experiment 1) and once within- ments the reminder bias was still present in the loss participants (Experiment 2). Experiment 2 further meas- framing condition. Therefore, participants were still ured metacognitive judgments and provided participants risk-averse and preferred to engage in offloading behav - with financial incentives to offload optimally. Participants ior even though reminders implied a loss. That is, despite were less biased toward using reminders when reminders reminders being framed in terms of losses participants were framed as losses instead of gains. This was also true still preferred the outcome that was more certain (using when participants received financial incentives for opti - reminders) over the uncertain and more variable out- mal offloading in Experiment 2. Further, the bias toward come (using internal memory). Hence, we observed a using reminders was higher the more underconfident preference shift but no preference reversal. This suggests participants were about their internal memory abilities, that, in the context of cognitive offloading, the framing of irrespective of the framing of reminders. The framing of choices in terms of losses does not loom large enough to reminders as well as participants’ internal metacogni- reverse the systematic preference for certain outcomes, tive bias were two independent factors, each providing thus leaving some amount of reminder bias. This finding unique predictive value regarding the optimality of par- may be due to domain-related characteristics of offload - ticipants’ offloading choices. ing situations which, by nature, could have impeded the In line with previous literature on decision-making reversal of risk preferences between framing conditions (Kahneman & Tversky, 1979), we observed that in situa- (for a review on moderating variables under gain and tions involving gains, individuals employed a risk-averse loss framing, see Piñon & Gambara, 2005). For example, offloading strategy in situations involving gains and as individuals are usually more sensitive for decisions under a result used more reminders, compared to situations risk, rather than uncertainty (Tversky & Fox, 1995). Thus, where offloading implied a loss. The experiments’ find - adding accuracy-based feedback for each individual, ings are thus consistent to studies demonstrating framing that is ensuring that the probabilities associated with the effects and reflection effects on decision-making in other respective strategies are known, may dissolve, at least, domains (e.g., Bless et al., 1998; Highhouse & Paese, 1996; some domain-specific constraints of offloading situa - Mather et al., 2012; for a review see Baucells & Villaís, tions (which one could argue would then lead to a further 2010; Kühberger, 1998; Piñon & Gambara, 2005; Steiger reduction of the reminder bias). & Kühberger, 2018; Tversky & Kahneman, 1981). How- A second explanation for individuals not being entirely ever, our study is the first to demonstrate effects of gain risk-seeking in the loss framing condition concerns other and loss framing in the context of cognitive offloading. influence factors that were still present in situations A possible explanation for the differences in offload - involving losses, providing a separate effect on the deci - ing strategies between framing conditions concerns sion to use reminders. Specifically, participants’ internal the certainty effect (Kahneman & Tversky, 1979). As metacognitive bias is one such additional factor influ - demonstrated in both experiments, participants in the encing participants’ offloading decisions (Boldt & Gil - gain condition were systematically biased toward using bert, 2019; Engeler & Gilbert, 2020). In Experiment 2, we reminders, thus preferring the option that provided more measured participants’ metacognitive judgments and we certainty regarding the outcome. This indicates that when observed that our participants were underconfident in the choice between reminders or internal memory is their internal memory capabilities and that this internal framed in terms of gains, individuals are risk-averse and metacognitive bias also contributed to the reminder bias, prefer the certain over the risky outcome, which is in line irrespective of the framing of reminders. Interestingly, with the certainty effect (Kahneman & Tversky, 1979). Engeler and Gilbert (2020) indicate that correcting par- In the loss condition, however, the systematic overuse of ticipants’ internal metacognitive judgments via positive reminders was reduced indicating that participants were feedback about their actual task performance may not more willing to take the risky outcome of potentially for- be sufficient to eliminate the reminder bias. That is, even getting a delayed intention and therefore losing a greater though participants no longer needed to be undercon- number of points by using their internal memory. This fident in their memory abilities, they still showed a sys - finding is consistent to the literature on decision-making tematic bias toward using reminders (Engeler & Gilbert, F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 15 of 18 2020). If one combines those findings with the results of Consistent with research on the original optimal our Experiment 2, which suggested that the framing of reminders task (Gilbert et al., 2020), participants were reminders influenced the reminder bias over and above consistently biased toward using reminders in the gain participants’ internal metacognitive bias, one could argue condition of both experiments, thus engaging in offload - that the framing of reminders and participants’ internal ing behavior more than optimal (Scarampi & Gilbert, metacognitive bias are two independent and additive fac- 2020). However, the reminder bias was still evident in tors influencing participants offloading behavior. the loss condition. With this first experimental demon - A third influence factor on the choice of using remind - stration that individuals deviate from optimal decision- ers is cognitive effort. When participants are provided making in situations where offloading involves possible with a performance-based financial incentive, they invest losses, we add to previous literature. Furthermore, as more cognitive effort resulting in an increased overall both experiments varied with respect to multiple factors task performance and a more optimal use of reminders (e.g., different samples; manipulation within-participants (Sachdeva & Gilbert, 2020). Despite providing a perfor- vs. between-participants; different scoring schemes; mance-based financial incentive in our second experi - see above, for further details), we demonstrate that the ment, participants still showed a significant (though systematic overuse of reminders constitutes a funda- reduced) reminder bias when reminders were framed in mental bias in human decision-making that occurs inde- terms of losses. The combination of a performance-based pendently of subtle variations in context (De Martino financial incentive with a loss framing was not sufficient et al., 2006; Hartley & Phelps, 2012; Mulder et al., 2012; to eliminate the reminder bias. This is consistent to the Tom et al., 2007). view that some decisions are rather difficult to debias, especially if relying on a nonlinear relationship between Implications the decision prospects (gain/benefit vs. loss/cost) and Offloading information onto external artifacts can signifi - the subjective values assigned to those prospects (Arkes, cantly help to fulfil future intentions, with participants 1991). Future research should thus continue investigat- consistently remembering more than 90% of inten- ing the joint––and hopefully debiasing––role of framing, tions using reminders in our experiments. This benefit metacognition, and cognitive effort on cognitive offload - of offloading on memory performance replicates results ing, with a specific focus on how to best achieve more found in previous studies (Engeler & Gilbert, 2020; Gil- optimal decisions in offloading situations. bert et al., 2020; Scarampi & Gilbert, 2020). However, Despite the overall reduction of the reminder bias using external devices in daily life does not always incur induced by loss framing, we also observed a rather con- a gain. For example, although offloading can help remem - sistent tendency of overusing reminders across par- ber intentions when memory resources become taxed, it ticipants. In particular, the correlation analysis between takes time and effort to set them up. Even though these individual reminder biases in the gain framing and loss individual costs of setting a reminder may be small, they framing conditions in Experiment 2 indicated that indi- mount to an unacceptable level if reminders were used vidual differences in participants’ offloading strate - for every one of the dozens, maybe hundreds, of things gies were consistent over the two framing conditions. one intends to do in an ordinary day. Other costs of using This may suggest that participants tendency to overuse reminders may only become apparent in the long run. reminders may be partially due to a preference for con- That is, apart from needing more time and effort, individ - sistency in performance, regardless of choice. That is, by uals miss out important cognitive exercise when setting having a general tendency to use reminders, participants up a reminder, possibly affecting cognitive function - are consistent and avoid the higher variability in accu- ing eventually (see Grinschgl et al., 2021b). In fact, the racy that occurs when using an internal memory strat- human brain is highly plastic (Green & Bavelier, 2008), egy. Therefore, participants may be more willing to avoid requiring exercise to establish and maintain proper func- variance in performance even when using their inter- tioning, especially at an older age (Buckner, 2004; Joubert nal memory may result in more correct responses and & Chainay, 2018; Morrison & Chein, 2011; Park & Bis- a higher financial reward. This explanation is consistent chof, 2013; Small, 2001). Therefore, when aiming toward with research proposing that individual’s preferences are an optimal offloading behavior, in real life one faces the risk-averse in terms of mental effort, opting for a fixed challenge how to deal with such costs and benefits of amount of effort rather than a variable amount (Apps using reminders (i.e., how harmful might it be to use the et al., 2015). This preference may be more prevalent when offloading strategy instead of training the brain and how the task demands are difficult, especially when one choice does this relate to the resulting performance). may require significantly more cognitive effort expendi - In our experiments, we demonstrated that while indi- ture, as in the current study. viduals were biased toward using reminders when facing Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 16 of 18 programmed the general framework of the optimal reminders task. MB the prospect of a gain, individuals’ offloading decisions adapted the programming code for Experiment 1. All authors contributed to tended to be more optimal in situations involving losses. the data analysis and were responsible for the interpretation of the results. LF Therefore, framing in the sense of highlighting either drafted the manuscript with the support of all authors. All authors approved the final version of the manuscript for submission. costs or benefits of offloading may also be a promising tool in achieving a more optimal offloading regarding Author information the relation of costs and benefits in daily life. Specifically, Preregistrations as well as data, code, and materials have been made publicly available at OSF; see https:// osf. io/ 8shkf/ (Experiment 1), https:// osf. io/ 8zvf6/ with using framing as a tool one may be able to develop (Experiment 2). Parts of this work were presented at the Cognitive Offloading interventions targeting the perspective individuals adopt Meeting in July 2021 hosted from the University College London and the TeaP in offloading contexts. That is, in situations where peo - 2022 (Conference of Experimental Psychologists) organized by the University of Cologne. This work entails values also presented in a separate project: ple are biased one way or the other (toward or away from https:// doi. org/ 10. 1177/ 17470 21820 970156 offloading), the framing of the instructions could be used to make peoples’ offloading decisions more optimal and Funding Open Access funding enabled and organized by Projekt DEAL. We acknowl‑ this approach could be used to create personalized inter- edge support by Open Access Publishing Fund of University of Tübingen for ventions based on each individual’s particular bias. For covering a part of the article processing charges. example, directing individuals’ focus on the costs that Availability of data and materials each strategy (using reminders or memory) has for their The datasets generated and analyzed during the current study as well as the daily life functioning may encourage them toward more materials are available in the OSF repository, https:// osf. io/ 8shkf/ (Experiment flexible adaptions of their offloading strategy. In a clinical 1), https:// osf. io/ 8zvf6/ (Experiment 2). context, for example, this may be a promising technique in establishing an optimal balance between functional Declarations and compensational rehabilitation for patients with neu- Ethics approval and consent to participate rological disease (see Thöne-Otto & Walther, 2008, for Ethical approval was received from the UCL Research Ethics Committee further details). (1584/003), and participants provided informed consent before participat‑ ing in the study for Experiment 2. Experiment 1 was conducted in accord‑ ance with the APA guidelines for research ethics, and participants provided informed consent before participating. Conclusion With the present research, we argue that the optimality Consent for publication with which individuals engage in offloading behavior can Not applicable. be shaped by the way offloading situations are framed. Competing interests When situations are framed in terms of gains, reminders The authors declare that they have no competing interests. are preferred, and when situations are framed in terms of Author details losses, individuals are less risk-averse and use reminders Department of Psychology, University of Tübingen, Tübingen, Germany. more optimally. Whereas previous attempts in achieving 2 Institute of Cognitive Neuroscience, University College London, London, UK. more optimal decisions in offloading situations have been Received: 21 October 2021 Accepted: 20 June 2022 following a mainly metacognitive approach (Engeler & Gilbert, 2020; Gilbert et al., 2020; Risko & Gilbert, 2016), the present study adds the concept of framing to this research area. We demonstrate that individuals’ offload - References ing strategies under gain and loss framing are affected Apps, M. A. J., Grima, L. L., Manohar, S., & Husain, M. (2015). The role of cogni‑ in line with the risk preferences predicted from previ- tive effort in subjective reward devaluation and risky decision‑making. Scientific Reports. https:// doi. org/ 10. 1038/ srep1 6880 ous literature on decision-making (Kahneman & Tversky, Arkes, H. R. (1991). Costs and benefits of judgment errors: Implications for 1979), leading to more optimal decisions when empha- debiasing. Psychological Bulletin, 110(3), 486–498. https:// doi. org/ 10. 1037/ sizing the costs (losses) rather than the benefits (gains) 0033‑ 2909. 110.3. 486 Ballard, D. H., Hayhoe, M. M., & Pelz, J. B. (1995). Memory representations in of offloading behavior. Thus, we claim that it is essential natural tasks. Journal of Cognitive Neuroscience, 7(1), 66–80. to move forward toward an inclusive theory of offload - Bates, D., Maechler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed‑ effects ing behavior, to expand our knowledge on how to best models using lme4. Journal of Statistical Software, 67(1), 1–48. Baucells, M., & Villasís, A. (2010). Stability of risk preferences and the reflection improve humans’ decision-making in offloading contexts. effect of prospect theory. Theory and Decision, 68(1–2), 193–211. https:// doi. org/ 10. 1007/ s11238‑ 009‑ 9153‑3 Acknowledgements Bless, H., Betsch, T., & Franzen, A. (1998). Framing the framing effect: The impact Not applicable. of context cues on solutions to the ‘Asian disease’ problem. European Journal of Social Psychology, 28(2), 287–291. https:// doi. org/ 10. 1002/ (SICI) Author contributions 1099‑ 0992(199803/ 04) 28:2% 3c287:: AID‑ EJSP8 61% 3e3.0. CO;2‑U LF, AF, MB, and FP developed the study concept, contributed to the study Boag, R. J., Strickland, L., Heathcote, A., Neal, A., & Loft, S. (2019). Cognitive design, and were responsible for testing and data collection of Experiment 1. control and capacity for prospective memory in complex dynamic CW and SG developed the study concept, contributed to the study design, and were responsible for testing and data collection of Experiment 2. SG F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 17 of 18 environments. Journal of Experimental Psychology: General, 148(12), Highhouse, S., & Paese, P. W. (1996). Problem domain and prospect frame: 2181–2206. https:// doi. org/ 10. 1037/ xge00 00599 Choice under opportunity versus threat. Personality and Social Psychology Boldt, A., & Gilbert, S. J. (2019). Confidence guides spontaneous cognitive Bulletin, 22(2), 124–132. https:// doi. org/ 10. 1177/ 01461 67296 222002 offloading. Cognitive Research: Principles and Implications, 4(1), 45. https:// Joubert, C., & Chainay, H. (2018). Aging brain: The effect of combined cognitive doi. org/ 10. 1186/ s41235‑ 019‑ 0195‑y and physical training on cognition as compared to cognitive and physical Buckner, R. L. (2004). Memory and executive function in aging and AD: training alone—a systematic review. Clinical Interventions in Aging, 13, Multiple factors that cause decline and reserve factors that compensate. 1267–1301. https:// doi. org/ 10. 2147/ CIA. S1653 99 Neuron, 44(1), 195–208. https:// doi. org/ 10. 1016/j. neuron. 2004. 09. 006 Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision Cherry, K. E., & LeCompte, D. C. (1999). Age and individual differences in under risk. Econometrica, 47(2), 263–291. https:// doi. org/ 10. 2307/ 19141 85 prospective memory. Psychology and Aging, 14, 60–76. https:// doi. org/ 10. Kirk, P. A., Robinson, O. J., & Gilbert, S. J. (2021). Trait anxiety does not correlate 1037/ 0882‑ 7974. 14.1. 60 with metacognitive confidence or reminder usage in a delayed inten‑ Crump, M. J. C., McDonnell, J. C., & Gureckis, T. M. (2013). Evaluating Amazon’s tions task. Quarterly Journal of Experimental Psychology, 74(4), 634–644. mechanical turk as a tool for experimental behavioral research. PLoS ONE, https:// doi. org/ 10. 1177/ 17470 21820 970156 8(3), e57410. https:// doi. org/ 10. 1371/ journ al. pone. 00574 10 Kliegel, M., McDaniel, M. A., & Einstein, G. O. (2000). Plan formation, retention, De Martino, B., Kumaran, D., Seymour, B., & Dolan, R. J. (2006). Frames, biases and execution in prospective memory: A new approach and age‑related and rational decision‑making in the human brain. Science, 313(5787), effects. Memory and Cognition, 28(6), 1041–1049. https:// doi. org/ 10. 3758/ 684–687. https:// doi. org/ 10. 1126/ scien ce. 11283 56BF032 09352 Diedenhofen, B., & Musch, J. (2015). Cocor: A comprehensive solution for the Kool, W., McGuire, J. T., Rosen, Z. B., & Botvinick, M. M. (2010). Decision making statistical comparison of correlations. PLoS ONE, 10(4), e0121945. https:// and the avoidance of cognitive demand. Journal of Experimental Psychol- doi. org/ 10. 1371/ journ al. pone. 01219 45 ogy: General, 139, 665–682. https:// doi. org/ 10. 1037/ a0020 198 Einstein, G. O., & McDaniel, M. A. (1990). Normal aging and prospective Kühberger, A. (1998). The influence of framing on risky decisions: A Meta‑anal‑ memory. Journal of Experimental Psychology: Learning, Memory, and Cogni- ysis. Organizational Behavior and Human Decision Processes, 75(1), 23–55. tion, 16(4), 717–726. https:// doi. org/ 10. 1037/ 0278‑ 7393. 16.4. 717https:// doi. org/ 10. 1006/ obhd. 1998. 2781 Ellis, J., Kvavilashvili, L., & Milne, A. (1999). Experimental tests of prospective Levin, I. P., Gaeth, G. J., Schreiber, J., & Lauriola, M. (2002). A new look at framing remembering: The influence of cue ‑ event frequency on performance. effects: Distribution of effect sizes, individual differences, and independ‑ British Journal of Psychology, 90(1), 9–23. https:// doi. org/ 10. 1348/ 00071 ence of types of effects. Organizational Behavior and Human Decision 26991 61233 Processes, 88(1), 411–429. https:// doi. org/ 10. 1006/ obhd. 2001. 2983 Engeler, N. C., & Gilbert, S. J. (2020). The effect of metacognitive training on Leys, C., Ley, C., Klein, O., Bernard, P., & Licata, L. (2013). Detecting outliers: Do confidence and strategic reminder setting. PLoS ONE, 15(10), e0240858. not use standard deviation around the mean, use absolute deviation https:// doi. org/ 10. 1371/ journ al. pone. 02408 58 around the median. Journal of Experimental Social Psychology, 49(4), Fagley, N. S. (1993). A note concerning reflection effects versus framing effects. 764–766. https:// doi. org/ 10. 1016/j. jesp. 2013. 03. 013 Psychological Bulletin, 113(3), 451–452. https:// doi. org/ 10. 1037/ 0033‑ 2909. Linares, D., & López‑Moliner, J. (2019). Quickpsy: Fits psychometric functions for 113.3. 451 multiple groups (version 0.1.5.1). [R package]. CRAN. https:// dlina res. org/ Faul, F., Erdfelder, E., Lang, A.‑ G., & Buchner, A. (2007). G*Power 3: A flexible quick psy. html. statistical power analysis program for the social, behavioral, and biomedi‑ Mather, M., Mazar, N., Gorlick, M. A., Lighthall, N. R., Burgeno, J., Schoeke, A., & cal sciences. Behavior Research Methods, 39(2), 175–191. https:// doi. org/ Ariely, D. (2012). Risk preferences and aging: The “certainty effect” in older 10. 3758/ BF031 93146 adults decision making. Psychology and Aging, 27(4), 801–816. https:// doi. Fisher, R. A. (1925). Statistical methods for research workers. Oliver and Boyd. org/ 10. 1037/ a0030 174 http:// psych class ics. yorku. ca/. Meyerhoff, H. S., Grinschgl, S., Papenmeier, F., & Gilbert, S. J. (2021). Individual Gilbert, S. J. (2015a). Strategic offloading of delayed intentions into the exter ‑ differences in cognitive offloading: A comparison of intention offloading, nal environment. The Quarterly Journal of Experimental Psychology, 68(5), pattern copy, and short‑term memory capacity. Cognitive Research: Princi- 971–992. https:// doi. org/ 10. 1080/ 17470 218. 2014. 972963 ples and Implications. https:// doi. org/ 10. 1186/ s41235‑ 021‑ 00298‑x Gilbert, S. J. (2015b). Strategic use of reminders: Influence of both domain‑ Morrison, A. B., & Chein, J. M. (2011). Does working memory training work? general and task‑specific metacognitive confidence, independent of The promise and challenges of enhancing cognition by training working objective memory ability. Consciousness and Cognition, 33, 245–260. memory. Psychonomic Bulletin & Review, 18(1), 46–60. https:// doi. org/ 10. https:// doi. org/ 10. 1016/j. concog. 2015. 01. 0063758/ s13423‑ 010‑ 0034‑0 Gilbert, S. J., Bird, A., Carpenter, J. M., Fleming, S. M., Sachdeva, C., & Tsai, P.‑ C. Mulder, M. J., Wagenmakers, E. J., Ratcliff, R., Boekel, W., & Forstmann, B. U. (2020). Optimal use of reminders: Metacognition, effort, and cognitive (2012). Bias in the brain: A diffusion model analysis of prior probability offloading. Journal of Experimental Psychology: General, 149, 501–517. and potential payoff. The Journal of Neuroscience, 32(7), 2335–2343. https:// doi. org/ 10. 1037/ xge00 00652https:// doi. org/ 10. 1523/ JNEUR OSCI. 4156‑ 11. 2012 Gray, W. D., Sims, C. R., Fu, W.‑ T., & Schoelles, M. J. (2006). The soft constraints Papenmeier, F. (2018). Powerbydesign: Power estimates for ANOVA designs (ver‑ hypothesis: A rational analysis approach to resource allocation for interac‑ sion 1.0.4) [R package]. CRAN. https:// cran.r‑ proje ct. org/ packa ge= power tive behavior. Psychological Review, 113(3), 461–482. https:// doi. org/ 10. bydes ign. 1037/ 0033‑ 295X. 113.3. 461 Park, D. C., & Bischof, G. N. (2013). The aging mind: Neuroplasticity in response Green, C. S., & Bavelier, D. (2008). Exercising your brain: A review of human to cognitive training. Dialogues in Clinical Neuroscience, 15(1), 109–119. brain plasticity and training‑induced learning. Psychology and Aging, Pearson, K., & Filon, L. N. G. (1898). Mathematical contributions to theory of 23(4), 692–701. https:// doi. org/ 10. 1037/ a0014 345 evolution: IV. On the probable errors of frequency constants and on the Grinschgl, S., Meyerhoff, H. S., & Papenmeier, F. (2020). Interface and interaction influence of random selection and correlation. Philosophical Transactions design: How mobile touch devices foster cognitive offloading. Comput- of the Royal Society of London, Series A, 191, 229–311. https:// doi. org/ 10. ers in Human Behavior, 108, 106317. https:// doi. org/ 10. 1016/j. chb. 2020. 1098/ rsta. 1898. 0007 106317 Piñon, A., & Gambara, H. (2005). A meta‑analytic review of framing effect: Risky, Grinschgl, S., Meyerhoff, H. S., Schwan, S., & Papenmeier, F. (2021a). From meta‑ attribute and goal framing. Psicothema, 17(2), 325–331. cognitive beliefs to strategy selection: Does fake performance feedback Risko, E. F., & Dunn, T. L. (2015). Storing information in‑the ‑ world: Metacogni‑ influence cognitive offloading. Psychological Research Psychologische tion and cognitive offloading in a short ‑term memory task. Consciousness Forschung. https:// doi. org/ 10. 1007/ s00426‑ 020‑ 01435‑9 and Cognition, 36, 61–74. https:// doi. org/ 10. 1016/j. concog. 2015. 05. 014 Grinschgl, S., Papenmeier, F., & Meyerhoff, H. S. (2021b). Consequences of Risko, E. F., & Gilbert, S. J. (2016). Cognitive offloading. Trends in Cognitive Sci- cognitive offloading: Boosting performance but diminishing memory. ences, 20(9), 676–688. https:// doi. org/ 10. 1016/j. tics. 2016. 07. 002 Quarterly Journal of Experimental Psychology, 74(9), 1477–1496. https:// doi. Sachdeva, C., & Gilbert, S. J. (2020). Excessive use of reminders: Metacognition org/ 10. 1177/ 17470 21821 10080 60 and effort ‑minimisation in cognitive offloading. Consciousness and Cogni- Hartley, C. A., & Phelps, E. A. (2012). Anxiety and decision‑making. Biological tion, 85, 103024. https:// doi. org/ 10. 1016/j. concog. 2020. 103024 Psychiatry, 72(2), 113–118. https:// doi. org/ 10. 1016/j. biops ych. 2011. 12. 027 Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 18 of 18 Scarampi, C., & Gilbert, S. J. (2020). The effect of recent reminder setting on subsequent strategy and performance in a prospective memory task. Memory, 28(5), 677–691. https:// doi. org/ 10. 1080/ 09658 211. 2020. 17649 74 Small, S. A. (2001). Age‑related memory decline: Current concepts and future directions. Archives of Neurology, 58(3), 360–364. https:// doi. org/ 10. 1001/ archn eur. 58.3. 360 Steiger, A., & Kühberger, A. (2018). A meta‑analytic re ‑appraisal of the framing effect. Zeitschrift Für Psychologie, 226(1), 45–55. https:// doi. org/ 10. 1027/ 2151‑ 2604/ a0003 21 Thöne‑ Otto, A. I. T., & Walther, K. (2008). Assessment and treatment of prospec‑ tive memory disorders in clinical practice. In M. Kliegel, M. A. McDaniel, & G. O. Einstein (Eds.), Prospective Memory: Cognitive, Neuroscience, Devel- opmental, and Applied Perspectives (pp. 321–345). Taylor & Francis Group/ Lawrence Erlbaum Associates. Tom, S. M., Fox, C. R., Trepel, C., & Poldrack, R. A. (2007). The neural basis of loss aversion in decision‑making under risk. Science, 315(5811), 515–518. https:// doi. org/ 10. 1126/ scien ce. 11342 39 Tversky, A., & Fox, C. R. (1995). Weighing risk and uncertainty. Psychological Review, 102(2), 269–283. https:// doi. org/ 10. 1037/ 0033‑ 295X. 102.2. 269 Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychol‑ ogy of choice. Science, 211(4481), 453–458. https:// doi. org/ 10. 1126/ scien ce. 74556 83 New York. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub‑ lished maps and institutional affiliations. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Cognitive Research:Principles and Implications Springer Journals http://www.deepdyve.com/lp/springer-journals/framing-cognitive-offloading-in-terms-of-gains-or-losses-achieving-a-Fu3euZCBIX

Loading next page...

References (52)

M Mather, N Mazar, MA Gorlick, NR Lighthall, J Burgeno, A Schoeke, D Ariely (2012)
Risk preferences and aging: The “certainty effect” in older adults decision making
Psychology and Aging, 27
HR Arkes (1991)
Costs and benefits of judgment errors: Implications for debiasing
Psychological Bulletin, 110
S Grinschgl, HS Meyerhoff, F Papenmeier (2020)
Interface and interaction design: How mobile touch devices foster cognitive offloading
Computers in Human Behavior, 108
C Scarampi, SJ Gilbert (2020)
The effect of recent reminder setting on subsequent strategy and performance in a prospective memory task
Memory, 28
M Kliegel, MA McDaniel, GO Einstein (2000)
Plan formation, retention, and execution in prospective memory: A new approach and age-related effects
Memory and Cognition, 28
GO Einstein, MA McDaniel (1990)
Normal aging and prospective memory
Journal of Experimental Psychology: Learning, Memory, and Cognition, 16
A Boldt, SJ Gilbert (2019)
Confidence guides spontaneous cognitive offloading
Cognitive Research: Principles and Implications, 4
D Bates, M Maechler, B Bolker, S Walker (2015)
Fitting linear mixed-effects models using lme4
Journal of Statistical Software, 67
SJ Gilbert, A Bird, JM Carpenter, SM Fleming, C Sachdeva, P-C Tsai (2020)
Optimal use of reminders: Metacognition, effort, and cognitive offloading
Journal of Experimental Psychology: General, 149
MAJ Apps, LL Grima, S Manohar, M Husain (2015)
The role of cognitive effort in subjective reward devaluation and risky decision-making
Scientific Reports
A Kühberger (1998)
The influence of framing on risky decisions: A Meta-analysis
Organizational Behavior and Human Decision Processes, 75
A Steiger, A Kühberger (2018)
A meta-analytic re-appraisal of the framing effect
Zeitschrift Für Psychologie, 226
S Grinschgl, F Papenmeier, HS Meyerhoff (2021)
Consequences of cognitive offloading: Boosting performance but diminishing memory
Quarterly Journal of Experimental Psychology, 74
CA Hartley, EA Phelps (2012)
Anxiety and decision-making
Biological Psychiatry, 72
SA Small (2001)
Age-related memory decline: Current concepts and future directions
Archives of Neurology, 58
PA Kirk, OJ Robinson, SJ Gilbert (2021)
Trait anxiety does not correlate with metacognitive confidence or reminder usage in a delayed intentions task
Quarterly Journal of Experimental Psychology, 74
EF Risko, TL Dunn (2015)
Storing information in-the-world: Metacognition and cognitive offloading in a short-term memory task
Consciousness and Cognition, 36
B De Martino, D Kumaran, B Seymour, RJ Dolan (2006)
Frames, biases and rational decision-making in the human brain
Science, 313
SJ Gilbert (2015)
Strategic offloading of delayed intentions into the external environment
The Quarterly Journal of Experimental Psychology, 68
EF Risko, SJ Gilbert (2016)
Cognitive offloading
Trends in Cognitive Sciences, 20
SM Tom, CR Fox, C Trepel, RA Poldrack (2007)
The neural basis of loss aversion in decision-making under risk
Science, 315
DH Ballard, MM Hayhoe, JB Pelz (1995)
Memory representations in natural tasks
Journal of Cognitive Neuroscience, 7
SJ Gilbert (2015)
Strategic use of reminders: Influence of both domain-general and task-specific metacognitive confidence, independent of objective memory ability
Consciousness and Cognition, 33
AB Morrison, JM Chein (2011)
Does working memory training work? The promise and challenges of enhancing cognition by training working memory
Psychonomic Bulletin & Review, 18
MJ Mulder, EJ Wagenmakers, R Ratcliff, W Boekel, BU Forstmann (2012)
Bias in the brain: A diffusion model analysis of prior probability and potential payoff
The Journal of Neuroscience, 32
HS Meyerhoff, S Grinschgl, F Papenmeier, SJ Gilbert (2021)
Individual differences in cognitive offloading: A comparison of intention offloading, pattern copy, and short-term memory capacity
Cognitive Research: Principles and Implications
WD Gray, CR Sims, W-T Fu, MJ Schoelles (2006)
The soft constraints hypothesis: A rational analysis approach to resource allocation for interactive behavior
Psychological Review, 113
B Diedenhofen, J Musch (2015)
Cocor: A comprehensive solution for the statistical comparison of correlations
PLoS ONE, 10
RL Buckner (2004)
Memory and executive function in aging and AD: Multiple factors that cause decline and reserve factors that compensate
Neuron, 44
AIT Thöne-Otto, K Walther, M Kliegel, MA McDaniel, GO Einstein (2008)
Assessment and treatment of prospective memory disorders in clinical practice
Prospective Memory: Cognitive, Neuroscience, Developmental, and Applied Perspectives
MJC Crump, JC McDonnell, TM Gureckis (2013)
Evaluating Amazon’s mechanical turk as a tool for experimental behavioral research
PLoS ONE, 8
K Pearson, LNG Filon (1898)
Mathematical contributions to theory of evolution: IV. On the probable errors of frequency constants and on the influence of random selection and correlation
Philosophical Transactions of the Royal Society of London, Series A, 191
D Kahneman, A Tversky (1979)
Prospect theory: An analysis of decision under risk
Econometrica, 47
C Sachdeva, SJ Gilbert (2020)
Excessive use of reminders: Metacognition and effort-minimisation in cognitive offloading
Consciousness and Cognition, 85
RJ Boag, L Strickland, A Heathcote, A Neal, S Loft (2019)
Cognitive control and capacity for prospective memory in complex dynamic environments
Journal of Experimental Psychology: General, 148
KE Cherry, DC LeCompte (1999)
Age and individual differences in prospective memory
Psychology and Aging, 14
A Tversky, D Kahneman (1981)
The framing of decisions and the psychology of choice
Science, 211
C Leys, C Ley, O Klein, P Bernard, L Licata (2013)
Detecting outliers: Do not use standard deviation around the mean, use absolute deviation around the median
Journal of Experimental Social Psychology, 49
M Baucells, A Villasís (2010)
Stability of risk preferences and the reflection effect of prospect theory
Theory and Decision, 68
S Highhouse, PW Paese (1996)
Problem domain and prospect frame: Choice under opportunity versus threat
Personality and Social Psychology Bulletin, 22
C Joubert, H Chainay (2018)
Aging brain: The effect of combined cognitive and physical training on cognition as compared to cognitive and physical training alone—a systematic review
Clinical Interventions in Aging, 13
J Ellis, L Kvavilashvili, A Milne (1999)
Experimental tests of prospective remembering: The influence of cue-event frequency on performance
British Journal of Psychology, 90
IP Levin, GJ Gaeth, J Schreiber, M Lauriola (2002)
A new look at framing effects: Distribution of effect sizes, individual differences, and independence of types of effects
Organizational Behavior and Human Decision Processes, 88
NS Fagley (1993)
A note concerning reflection effects versus framing effects
Psychological Bulletin, 113
A Piñon, H Gambara (2005)
A meta-analytic review of framing effect: Risky, attribute and goal framing
Psicothema, 17
DC Park, GN Bischof (2013)
The aging mind: Neuroplasticity in response to cognitive training
Dialogues in Clinical Neuroscience, 15
A Tversky, CR Fox (1995)
Weighing risk and uncertainty
Psychological Review, 102
H Bless, T Betsch, A Franzen (1998)
Framing the framing effect: The impact of context cues on solutions to the ‘Asian disease’ problem
European Journal of Social Psychology, 28
CS Green, D Bavelier (2008)
Exercising your brain: A review of human brain plasticity and training-induced learning
Psychology and Aging, 23
F Faul, E Erdfelder, A-G Lang, A Buchner (2007)
G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences
Behavior Research Methods, 39
W Kool, JT McGuire, ZB Rosen, MM Botvinick (2010)
Decision making and the avoidance of cognitive demand
Journal of Experimental Psychology: General, 139
NC Engeler, SJ Gilbert (2020)
The effect of metacognitive training on confidence and strategic reminder setting
PLoS ONE, 15

Publisher: Springer Journals
Copyright: Copyright © The Author(s) 2022
eISSN: 2365-7464
DOI: 10.1186/s41235-022-00416-3
Publisher site: See Article on Publisher Site

Abstract

Nowadays individuals can readily set reminders to offload intentions onto external resources, such as smartphone alerts, rather than using internal memory. Individuals tend to be biased, setting more reminders than would be optimal. We address the question whether the reminder bias depends on offloading scenarios being framed as either gains or losses, both between‑participants (Experiment 1) and within‑participants (Experiment 2). In both experi‑ ments, framing of reminders in terms of gains resulted in participants employing a risk‑averse strategy and using more reminders than would be optimal. Importantly, however, participants used reminders more optimally and were more willing to choose the risk‑seeking option of remembering internally when reminders implied a loss. Based on metacognitive measures in Experiment 2, the reminder bias increased the more underconfident participants were about their memory abilities in both framing scenarios. Framing did not alter this relationship between erroneous metacognitive underconfidence and reminder bias but provides an additional influence. We conclude that empha‑ sizing the losses (costs) associated with external reminders helps in achieving more optimal decisions in offloading situations, and that in addition to cognitive effort and metacognitive judgments, framing needs to be considered in improving individuals’ offloading behavior. Keywords: Framing, Offloading, Reminders, Metacognition, Prospective memory Public significance statement of the combination of their internal memory resources Faced with the choice of remembering with inter- and the external aids. Therefore, our findings may help to nal memory or using external reminders (e.g., smart- optimize the use of memory aids, that is to compensate phone apps, calendars), individuals generally tend to use for memory limitations in everyday life whenever neces- reminders more than optimal. Our results suggest that sary but at the same time to not neglect internal cognitive emphasizing either the gains (benefits) or losses (costs) resources. associated with external reminders influences individu - als’ decision between using internal memory or exter- Introduction nal reminders. Emphasizing the losses (costs) associated Suppose you have to make an appointment for 4 pm with external reminders led individuals to rely less on tomorrow. Part of everyday life requires remember- external reminders and thus to make more optimal use ing delayed intentions that are fulfilled in the future and stored in prospective memory (Einstein & McDaniel, 1990). However, the capacity of prospective memory is limited (e.g., Cherry & LeCompte, 1999), leading to fre- *Correspondence: frank.papenmeier@uni‑tuebingen.de quent failures in remembering delayed intentions, which Department of Psychology, University of Tübingen, Tübingen, Germany can interfere with functioning in everyday life (Boag Full list of author information is available at the end of the article © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 2 of 18 et al., 2019; Ellis et al., 1999; Kliegel et al., 2000). As tech- intentions maintained over a typical day, including those nology has become commonplace nowadays (e.g., smart- that are trivial or highly practiced such as remembering phones or digital watches), individuals usually have the to go to work, brush one’s teeth, eat, or sleep. An addi- option to offload information onto external devices to tional possible cost of external reminders is that they may help them remember intentions. Using external artifacts prevent an opportunity to strengthen internal cognitive to reduce cognitive demand is known as cognitive off - skills (though see Scarampi & Gilbert, 2020). loading (Risko & Gilbert, 2016). For example, users can readily set up reminders on their smartphones to remind Optimal use of offloading them of their appointment. Considering the decision between relying on offloading Multiple views have been proposed attempting to tools versus internal memory as a tradeoff between costs explain individuals’ choice between relying either on and benefits raises the question of whether there might internal cognitive resources or external aids when solv- be an optimal solution for this tradeoff. With optimality, ing tasks. A minimal memory view proposes that humans we refer to a decision that perfectly balances the costs have the tendency to store information externally when- (e.g., time and effort related to offloading) and the ben - ever possible (Ballard et al., 1995). This is consistent with efits of offloading (e.g., reduced cognitive demand and findings suggesting that individuals aim to avoid actions increased accuracy). Accordingly, we refer to decisions associated with cognitive demand (Kool et al., 2010). as optimal if they maximize the benefits while minimiz - However, this would also imply that individuals should ing the costs of offloading, that is, relying on offloading always use offloading tools regardless of other factors, tools as often as necessary but as seldom as possible. In which is clearly not the case. In contrast, individuals’ turn, we define biases as deviations from this normative choice between either offloading or relying on internal decision-making model. resources is determined by multiple factors, such as To quantify optimality and bias in the context of cogni- memory capacity (Meyerhoff et al., 2021), memory load tive offloading research, Gilbert et al. (2020) introduced (Gilbert, 2015a), monetary reward (Sachdeva & Gilbert, the so-called optimal reminders task. In this task, par- 2020), metacognitive judgments regarding one’s own ticipants are instructed to drag circles to the bottom of internal abilities (Boldt & Gilbert, 2019; Gilbert, 2015b; a square box in ascending numerical order. Sometimes, Risko & Gilbert, 2016; but see Grinschgl et al., 2021a), special circles that were briefly filled with a different color or the interaction and interface design of offloading appeared, and participants had to remember the delayed tools (Grinschgl et al., 2020). Other influence factors on intention of dragging those circles to the correspond- offloading are more generic, such as context (e.g., time ingly-colored border when it was their turn. In order for frame, device) or personal preferences (e.g., personality, participants to fulfill these intentions, they introduced consequences of a missed appointment). them with two strategies: relying on internal memory or The decision to engage in offloading behavior is also setting external reminders. Whereas in some trials, par- affected by cost–benefit considerations (Gray et al., ticipants were forced to use either internal memory or 2006). The most obvious benefit of using offloading tools external reminders, other trials gave participants a free rather than relying on internal resources is that with off - choice between scoring a maximum amount of points loading remembering the offloaded information is nearly using memory or a lesser amount of points using remind- guaranteed (Risko & Gilbert, 2016). This accuracy-related ers. The number of points gained when using reminders benefit of offloading accounts not only for daily life, such was manipulated across trials. Based on participants’ as remembering the items to buy in the grocery with performance in the forced trials, Gilbert et al. (2020) cal- either a shopping list or internal memory. It also accounts culated a normative optimal points value at which par- for well-established offloading tasks, such as the inten - ticipants should switch from using reminders to using tion offloading task (e.g., Gilbert et al., 2020) or the task memory. Based on actual choice behavior, they evalu- of remembering information with the support of writing ated participants’ bias. They observed that participants it down (e.g., Risko & Dunn, 2015). In addition, outsourc- did not show optimal choices. Instead, participants chose ing cognitive demand onto external tools can sometimes reminders more than optimal, thus demonstrating a bias be considered less effortful compared to remembering toward reminders––the reminder bias (Gilbert et al., internally (Ballard et al., 1995; Sachdeva & Gilbert, 2020). 2020). External reminders also incur costs, however. In eve- The size of this reminder bias for delayed intentions ryday life, these costs include the time and effort of set - can be influenced by individuals’ metacognitive judg - ting them up, and the interruptions they can cause. These ments regarding the subjective perception of their costs may be individually minimal. But they would mount internal abilities (Boldt & Gilbert, 2019; Risko & Gil- to an unacceptable level when applied to the multitude of bert, 2016). Reminder use was predicted by individuals’ F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 3 of 18 erroneous underconfidence in their memory abilities outcome (win money vs. loose money), but it occurs also (Gilbert, 2015a; Gilbert et al., 2020). Specifically, the if the same outcome is phrased to appear to involve either reminder bias was higher among those individuals who gains or losses, the so-called framing effect (e.g., Bless were underconfident about their own memory (Enge - et al., 1998; Fagley, 1993; Highhouse & Paese, 1996; for ler & Gilbert, 2020). However, when Gilbert et al. (2020) a review, see Kühberger, 1998; Piñon & Gambara, 2005; corrected participants’ confidence via positive feedback, Steiger & Kühberger, 2018; Tversky & Kahneman, 1981). thus aiming at debiasing participant’s offloading choices For example, under a gain framing, individuals would by making them less underconfident, participants were prefer the risk-averse option of winning $300 with cer- less biased toward using reminders but still offloaded tainty over the risky option of winning either $1000 with more than would be optimal. In a similar vein, provid- 30% probability or $0 with 70% probability. If the same ing participants with a financial incentive based on over - outcome (average profit of $300) is framed under losses, all performance reduced the reminder bias, thus causing individuals show a shift to a risk-seeking strategy. That is, participants to use reminders more optimally (Sachdeva if one tells individuals that they have an initial $1000 and & Gilbert, 2020), likely because participants were will- that they could choose between the option of losing $700 ing to invest more cognitive effort. However, despite the with certainty and the option of losing either $0 with 30% reduction in the reminder bias, this was again not enough probability or $1000 with 70% probability, individuals to debias participant’s offloading choices. Thus, if used as would prefer the later (risky) option. a tool to achieve a more optimal use of reminders, both Applying those findings from the literature on deci - approaches were only partially effective in reducing the sion-making (Kahneman & Tversky, 1979) to cognitive reminder bias, suggesting that there must be further offloading makes evident that the previously reported barriers that first have to be overcome in order for the reminder bias (Gilbert et al., 2020) could also be seen reminder bias to be fully eliminated. as individuals employing a risk-averse choice strategy. Within previous research on the reminder bias (Engeler Offloading vs. internal memory: decision & Gilbert, 2020; Gilbert et al., 2020; Sachdeva & Gilbert, under uncertainty 2020), the choice between using reminders or internal We take a novel perspective on individuals’ offloading memory was framed in terms of gains, such as earning decisions. Considering the typically achieved perfor- some points using reminders and earning a maximum mance when using an offloading strategy compared to reward using memory. Applying the certainty effect to internal memory, it turns out that performance is usually cognitive offloading, one would thus expect that individ - very high or even near perfect with offloading compared uals prefer outcomes that are near-certain (using an off - to a much more unreliable performance when relying on loading strategy) over outcomes that are more uncertain the internal memory. u Th s, when choosing between off - (using internal memory), thus resembling the reminder loading or internal memory, individuals decide between bias. achieving a relatively certain outcome when using the off - Given the idea that the choice between cognitive off - loading strategy and an uncertain or risky outcome when loading and internal memory represents a decision under using internal memory. uncertainty, a switch from gain framing to loss framing When making decisions under risk, individuals typi- should also cause a switch from risk-averse choice strate- cally prefer certain over uncertain choices, also known gies to risk-seeking choice strategies within the context as the certainty effect (Kahneman & Tversky, 1979). For of cognitive offloading. Thus, with the aim of using the example, individuals would prefer winning $3000 with framing of reminders as a means of achieving a more certainty over winning $4000 with 80% probability, even optimal use of reminders, reversing the framing from though the expected outcome is higher in the second gains to losses might reduce, eliminate, or even reverse variant. Interestingly, this bias reverses when the choice the reminder bias. With the present research, we inves- options are presented as losses rather than gains, also tigated whether this shift in the reminder bias could be known as the reflection effect (Kahneman & Tversky, observed. 1979). For example, individuals would prefer losing $4000 with an 80% probability over losing $3000 with certainty, Experiment 1 even though the expected loss is lower in the second vari- Our first experiment used the optimal reminders task ant. That is, switching from gains to losses also causes a (Gilbert et al., 2020) and expanded it by implementing a shift from risk-averse choice strategies to risk-seeking between-participants manipulation regarding the fram- choice strategies (e.g., Baucells & Villaís, 2010; Kahneman ing of reminders. We framed reminders as gains, just as & Tversky, 1979; Mather et al., 2012). This shift in choice previous research, or as losses, which had not been done strategies occurs not only if there is an actual change in before. We had two key hypotheses. In the gain framing Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 4 of 18 condition, we expected to replicate previous findings on cognitive offloading, thus showing a reminder bias (Gil - bert et al., 2020). For the loss condition, we expected a reduction of the reminder bias. This experiment was pre - registered; see https:// osf. io/ rcu8v. Method Participants We recruited the participants using the student mailing list of the University of Tübingen and online posts shared in non-commercial Facebook and WhatsApp groups. Following exclusions (see below), our sample consisted of 141 participants (94 reporting their gender as male, 32 as female, and 2 as other). Participants had a mean age of 26.30 years (SD = 8.90, range = 18–64). The experi - ment took approximately 45 min, for which participants received the opportunity to win one of six €10 gift cards from a local bookstore or train company. We conducted Fig. 1 Illustration of the optimal reminders task. Note Example trial of the optimal reminders task. a Participants were instructed to drag this experiment in accordance with the APA guidelines circles to the bottom edge of a box in sequential order. Each time a for research ethics, and participants provided informed circle was moved onto an edge, it disappeared from the screen and consent before participating. the next circle in sequence emerged; b Sometimes, new circles were We performed a power analysis based on the results initially highlighted in a different color, indicating a delayed intention observed in the unadvised group of Experiment 2 in Gil- to drag the special circle to the same‑ colored edge when reached in sequence; c A special circle’s color faded back to yellow two seconds bert et al. (2020), using the R package powerbydesign after appearance. d If permitted, participants set reminders by (Papenmeier, 2018). To achieve a power of 80% for the instantly dragging the special circles near their intended edge when investigated interaction effect (assumption of reminder they emerged on the screen; e Participants carried on with the task of bias under gain framing and no bias under loss framing; dragging circles to the bottom of the box; f After dragging the circles see the script containing the power simulation for details: in sequential order, they could then execute the delayed intention to drag the special circle to its intended location https:// osf. io/ xqt8j), we required a sample size of 136 participants. We stopped data collection after two weeks, with the study slightly overshooting the targeted sample size (N = 141, 9 excluded) at this point in time. toward using or their own memory, or (c) optimally calibrated. On each trial, participants used their computer mouse Optimal reminders task to sequentially drag 25 numbered circles to the bottom We modified the optimal reminders task used by Gilbert of a box (Fig. 1). Up to six circles were visible a time, et al. (2020). In this task, participants can choose (a) to and each time a circle was removed from the box, it was remember intentions using internal memory, which replaced with a new one (e.g., after dragging ‘1’ to the leads to a maximum reward for each remembered item, bottom, a new circle labeled ‘7’ appeared in its place). The or (b) they can set external reminders, which leads to a left, top, and right edges of the box were colored blue, smaller reward that varies from trial to trial. This para - orange, and purple, respectively. Occasionally, new cir- digm allowed us to examine not only the frequency of cles appeared initially in one of these colors before fad- reminder-setting but also its optimality. For example, ing to yellow after 2 s. This was an instruction to form suppose an individual’s accuracy is 55% when using their a delayed intention to drag these ‘special’ circles to the own memory and 100% when using reminders. If they are corresponding edge of the box. For example, if a spe- given a choice between earning 10 points per item using cial circle (e.g., 7) initially appeared as blue, participants their own memory or 5 points per item using reminders, needed to remember this instruction while they dragged it is optimal to use the internal memory strategy. But if circles 2 to 6 to the bottom of the box (by which time they are offered 6 points per item when using reminders, the special circle had faded to yellow). They could then it is optimal to select this strategy instead. By compar- execute the intention to drag 7 to the left. Within each ing participants’ choices with the optimal strategy, this trial consisting of 25 circles, 10 special circles were pre- paradigm can be used to calculate whether individuals sented. These circles appeared between the 7th and 25th are (a) biased toward using external reminders, (b) biased F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 5 of 18 circle in the sequence. (The initial 6 circles were already the width/height of the box, and all circles were initially on screen at the beginning of the trial, so they could not placed so that they fall within a central portion of the box act as targets.) These 19 possible target positions were with dimensions sized at 56% of the total width/height, split into 10 adjacent bins (9 of which had a length of two so that no circles were adjacent to any of the edges of the and one of which, placed randomly in the sequence, had box at the beginning of the trial. length one). One target was then placed randomly within each of these bins. As a result of the multiple concur- Procedure rent intentions, participants were unlikely to remember After providing informed consent, the computer ran- all if they relied on internal memory alone. Alternatively, domly assigned participants to one framing condition, if they used reminders, they could offload the inten - with the assignment resulting in 63 and 78 participants tions by immediately dragging special circles next to for the gain and loss framing, respectively. Participants the instructed edge when they first appeared (e.g., drag - then completed the six practice trials: In the first trial, ging a blue 7 toward the left edge of the box as soon as participants dragged the circles in sequence to the bot- it appeared rather than waiting for it to fade to yellow tom of the square (8 circles in total). They further prac - first). The location of the special circle then acted as a ticed by dragging one special circle to the instructed reminder when the participant reached this number in edge in the second practice trial (8 circles in total). They the sequence. An everyday analogy would be leaving an repeated this practice trial and were not allowed to con- object by the front door so that you remember it when tinue the experiment until they responded correctly to leaving the house the next day. this special circle. They then continued with two con - Participants alternated between ‘forced’ and ‘choice’ secutive practice trials of actual length (10 special circles trials. On forced trials, they had to use either their own out of 25 circles in total). Following this, they were made memory (‘forced-internal’) or reminders (‘forced-exter- aware of the ability to use reminders in this task. They nal’). On choice trials, participants decided between practiced again, but on the forced-external trial type. This earning 10 points per remembered item (using their own time, they needed to respond correctly to at least 8 out of memory), or a smaller number of points between 1 and 10 special circles in order to continue the experiment. In 9 (using external reminders). We calculated the optimal line with previous studies (Gilbert, 2015b; Gilbert, et al., strategy based on performance on the forced trials, then 2020), we used the exact value of eight correct responses compared this with their actual decisions on the choice to ensure that participants were able to achieve at least trials. The experiment was split into two conditions (gain 80% accuracy with using the external strategy. After per- and loss), with participants being assigned randomly to forming one additional forced-external practice trial, one condition. In both conditions, participants were participants were instructed about the upcoming forced given 0 points at the beginning of the experiment. Fur- and choice trials and that their task was to gain as many ther, participants in the gain condition chose between points as possible (lose as few points as possible). receiving 10 points for each remembered special circle or During the main experiment, participants performed a smaller number of points (1–9) to use reminders (see a total of 17 trials. On odd-numbered trials, participants Fig. 2). For each missed special circle, participants gained were given a free choice between using internal memory zero points. This matches the version of the task used in (gain 10 points/lose 0 points; according to framing con- previous research (Gilbert et al., 2020). In the loss condi- dition) or reminders (gain/lose 1–9 points per special tion, participants lost 10 points for each missed special circle, presented in random order). On even-numbered circle, and they had the choice between (a) using their trials, participants alternated between the forced-exter- own memory and keeping all their previously scored nal and forced-internal trials, with the starting trial type points (losing 0) each time they correctly remembered (external or internal) randomized between participants. special circles or (b) using reminders and scoring minus The trial number was set to the exact number of 17 (9 points every time they remembered (− 9 to − 1). All free, 8 forced) trials to be consistent with previous ver- instructions were presented in German. sions of the optimal reminders task (Gilbert et al., 2020). After completion, participants were given the opportu- Apparatus nity to enter the prize draw on SoSci Survey. Participants completed the task via their computer’s web browser. Participation was only permitted if the browser Reward window had dimensions of at least 500 × 500 pixels. The Participants were told that they were scoring points, square box containing the circles was sized at 80% of with the prospect of earning (losing) up to 1700 points the horizontal or vertical extent of the browser window, in the gain (loss) condition. Therefore, the earnings whichever was smaller. Each circle had a radius of 5.5% of could range between 0 and 1700 points in the gain Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 6 of 18 Fig. 2 Example instructions for the free‑ choice trials. Note Prior to the start of each free‑ choice trial, participants were given the choice to either rely on their internal memory or to set external reminders. a In the gain condition, participants would score 10 points for each special circle they remembered if they relied on their memory or a smaller number of points (1–9) if they selected to set reminders. b In the loss condition, if participants relied on their memory, they would lose 0 points for each special circle they remembered. If they chose to use reminders, however, they would score minus points (− 9 to − 1). For both framing conditions, a sequence of 25 circles was presented in each trial, with 10 of them acting as special circles condition, and between minus 1700 points and 0 points • Forced-internal accuracy (ACC ). This is the mean FI in the loss condition. The experiment was promoted by target accuracy (proportion of special circles cor- offering participants the chance to win one of six 10€ rectly dragged to the instructed location) on forced- gift cards for taking part in the experiment. internal trials. • Forced-external accuracy (ACC ). This is the mean FE target accuracy (proportion of special circles cor- rectly dragged to the instructed location) on forced- Design external trials. The experiment employed a 2 (framing condition: gain × • Optimal indifference point (OIP). For choice tri - vs. loss) 2 (indifference point: optimal vs. actual) als, this is the value for special circles offered with design with framing as a between-participants variable, reminders at which an unbiased individual should and we defined five variables of interest: F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 7 of 18 be indifferent between the two options, based on the it earned them fewer points, indicating random strat- accuracy in the forced-internal and forced-external egy selection; n = 8); and (e) their reminder bias score trials (ACC and ACC ). As in Gilbert et al. (2020), exceeded 2.5 standard deviations from the group mean FI FE this was calculated as (n = 1). The data are publicly available (https:// osf. io/ 8shkf/). OIP × ACC = 10 × ACC (1) FE FI Transparency and openness If the OIP was less than 1 or greater than 9, it was For the study’s entire research report, we communi- set to the relevant lower or upper bound. This was cate our methodical and statistical approach including so that the potential values of the OIP would match sample size estimation, data exclusion, experimental the potential values of the point at which they were manipulation, and measures of interest. All hypotheses, actually indifferent, which was bound by their experimental methods, and planned analyses were pre- choices for values 1 to 9. registered before data collection. All analyses were run in R, and information on the R environment (including package versions) used for the analyses is given in the • Actual indifference point (AIP). This is the esti - analysis script. mated point for choice trials at which participants were actually indifferent to the two strategy options. As in Gilbert et al. (2020), this was calculated by fit - Results ting a sigmoid curve to the strategy choices (0 = own Accuracy memory; 1 = reminders) across the nine special val- Participants were able to remember almost two-thirds of ues (1–9), using the quickpsy function from the R the special circles using their memory in forced-internal package quickpsy (Linares & López-Moliner, 2019) trials, but nearly all of them using reminders in forced- bounded to the range 1 to 9 (see the analysis script external trials (see Fig. 3a). This data is consistent to on OSF; https:// osf. io/ qsfmy/). Based on this curve, the accuracy data found in Gilbert et al. (2020). As an we were able to estimate the point associated with exploratory analysis, we submitted the accuracy data to 50% probability of choosing either strategy, which is a 2 (condition: forced-internal vs. forced-external; within) the AIP. × 2 (framing: gain vs. loss; between) mixed analysis of • Reminder bias. This is defined as OIP–AIP, which variance (ANOVA). Whereas the main effect of condition will yield a positive value for a participant biased was significant, F(1, 139) = 641.22, p < 0.001, η = 0.82, toward using more reminders than would be optimal, neither the main effect of framing, F(1, 139) = 0.03, and a negative value for a participant biased toward p = 0.856, η < 0.01, nor the interaction of framing using fewer reminders than would be optimal. and condition, F(1, 139) = 0.02, p = 0.899, η < 0.01, was significant. Thus, although participants showed a Each of the previous five measures was calculated sepa - higher task accuracy when using reminders, their over- rately for the gain and loss condition, and to compare all task performance was not affected by the framing of AIP and OIP between framing conditions, calculation of reminders. indifference points was performed after transforming the minus points from the loss framing condition by adding Reminder bias As defined in our preregistration, we investigated the influence of framing on the reminder bias (defined as OIP Exclusion criteria minus AIP) by submitting the reminder bias scores to a In accordance with our preregistration, we excluded 2 (framing condition: gain vs. loss; between) × 2 (indif- participants if (a) their accuracy in forced-internal tri- ference point: OIP vs. AIP; within) mixed ANOVA (see als (averaged across gain and loss conditions) was lower Fig. 3a). There was a significant main effect of indif - than 10% (n = 0); (b) accuracy in the forced-external tri- ference point, F(1, 139) = 52.17, p < 0.001, η = 0.27, als was lower than 70% (averaged across gain and loss and a non-significant main effect of framing condition, conditions; n = 0); (c) accuracy in the forced-internal tri- F(1, 139) = 3.45, p = 0.065, η = 0.02. Importantly, the als was higher than in the forced-external trials in either interaction of framing and indifference point was sig - condition (n = 0); (d) there was a negative point biserial nificant, F(1, 139) = 10.05, p = 0.002, η = 0.07, with the correlation between points offered for correct responses reminder bias scores, on average, 1.19 points greater in on each trial using reminders (1–9) and choice of strat- the gain (M = 2.00, SD = 2.45) compared to loss condi- egy (0 = own memory, 1 = reminders; this excluded par- tion (M = 0.81, SD = 2.00). We further analyzed whether ticipants who were more likely to set reminders when the reminder bias scores were significantly greater than Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 8 of 18 Fig. 3 Results on Offloading Behavior from Experiment 1. Note a In the left panel, mean accuracy for the forced‑internal and forced‑ external trials in the gain and loss framing conditions. Mean optimal (OIPs) and actual indifference points (AIPs) as a function of framing condition in the right panel. The line pattern illustrates the size of the reminder bias defined as OIP minus AIP. b Correlation between actual and optimal indifference points for the gain (left panel) and loss (right panel) framing condition. Error bars represent standard errors zero with preregistered one-tailed paired t tests (OIPs Fisher’s (1925) z = − 2.53, p = 0.011. That is, despite par - vs AIPs). Although framing reminders in terms of losses ticipants having a systematic bias toward using remind- instead of gains resulted in a significant reduction of the ers, those who derived the most benefit from reminders reminder bias, the reminder bias was still significant both (i.e., low OIP) were also most likely to use them (i.e., in the gain, t(62) = 6.49, p < 0.001, d = 0.82, and loss con- low AIP) and, although evident in both conditions, this dition, t(77) = 3.60, p < 0.001, d = 0.41. That is, although relationship was more pronounced when offloading was participants offloaded more optimally under loss framing associated with a loss (see Fig. 3). than under gain framing, participants still offloaded more than optimal also under loss framing. Discussion We ran an exploratory analysis on the relationship Consistent with our hypotheses, we observed a strong between the OIPs and AIPs. To achieve this, we calculated reminder bias in the gain condition of Experiment 1. That a Pearson’s product–moment correlation on the indiffer - is, participants offloaded more than optimal, replicat - ence points for each framing condition. We further tested ing previous findings (Gilbert et al., 2020). In addition, for the difference between correlations by using the R we showed for the first time that the framing of remind - package cocor (test of significance for independent cor - ers in terms of gains or losses shapes the reminder bias. relations; Diedenhofen & Musch, 2015). There was a sig - That is, the reminder bias was largely reduced under loss nificant positive correlation between the OIPs and AIPs framing. This is in line with the literature on decision- in both conditions (gain: r = 0.29, p = 0.019; loss: r = 0.63, making (Kahneman & Tversky, 1979), as it suggests that p < 0.001; see Fig. 3b), with the correlation being signifi - while individuals employ a risk-averse strategy in situ- cantly higher in the loss compared to the gain condition, ations involving gains, they are more inclined to take a F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 9 of 18 risk in situations involving losses. However, it is to note studies in this meta-analysis, at least 49 participants were that despite participants being more risk-seeking when required for 80% power ( α = 0.05, two-tailed paired t facing the prospect of a loss, they still deviated from opti- test). This experiment also provided data for a separate mal offloading behavior in the loss framing condition, unrelated project, which was reported elsewhere (Kirk thus using reminders more than optimal. This finding is et al., 2021). Therefore, we were aiming for a higher sam - consistent with the view that also other factors influence ple size (i.e., N = 300), as this was the intended sample the reminder bias, such as participants’ underconfidence size for the other project. As in earlier studies (Gilbert, in their memory abilities (Gilbert et al., 2020) and their 2015a, 2015b), participation was restricted to volun- invested cognitive effort (Sachdeva & Gilbert, 2020). We teers aged at least 18 years, located in the USA. We also addressed these points in Experiment 2. restricted inclusion to participants with a minimum of 90% Mechanical Turk approval rate. Participation took Experiment 2 approximately 45 min, for which participants were guar- In Experiment 2, we again manipulated the framing of anteed a base payment of $2, plus an additional bonus reminders as gains or losses, while participants per- of up to $8.67 depending on their task performance. formed the optimal reminders task. In addition, we asked Participants had a mean age of 37.81 years (SD = 10.97, the participants to make metacognitive judgments at the range = 21–72); 190 reported their gender as male, 108 beginning of the experiment. This allowed us to assess as female, and 2 as other. Ethical approval was received their overconfidence or underconfidence in carrying from the UCL Research Ethics Committee (1584/003) out the task (Gilbert et al., 2020). Further, we rewarded and participants provided informed consent before par- participants depending on their task performance. This ticipating in the study. acted as a financial incentive to increase both the cog - nitive effort that participants invest while performing Optimal reminders task the task and the optimality of their offloading choices We used the same modified version of the optimal (Sachdeva & Gilbert, 2020). Furthermore, we manipu- reminders task as in Experiment 1, with small adapta- lated framing within-participants instead of between- tions. First, Experiment 2 used an English-speaking participants in order to account for potential individual MTurk sample, and thus, all items were presented in differences in framing effects (see Levin et al., 2002). As English. Second, we included metacognitive accuracy in Experiment 1, we expected to observe a reminder bias judgments in between the practice trials. That is, both in the gain framing condition, but a reduction (or even after practicing trials on the forced-internal and the reversal) of this bias in the condition where reminders forced-external type, participants provided a measure implied a loss. This experiment was preregistered; see of how confident they were at their ability to perform https:// osf. io/ 8zvf6/. the task (see Fig. 4). Third, framing was implemented as a within-participants variable. This allowed us to test Method whether results observed in Experiment 1 can also be Participants observed when comparisons between framing conditions Participants were recruited from the Amazon Mechani- are made within, rather than between individuals. To cal Turk website (http:// www. mturk. com), an online do this, the task was split into two blocks: gain and loss marketplace in which participants receive payment for (with the order of blocks being counterbalanced between completion of web-based tasks (Crump et al., 2013). participants). Sample size was estimated performing a power analysis Finally, we did not assign negative values to loss points. with G*Power 3.1 (Faul et al., 2007). The power analysis That is, while instructions of the gain condition stayed was based on the meta-analysis of Kühberger (1998), as the same, during the loss condition participants received to our knowledge there was no previous study that has the maximum of points available for this block (100 investigated the effect of gain and loss framing within- points per trial) before the beginning of the block. They participants in the context of cognitive offloading. With were presented with the choice between either using a Cohen’s d effect size of 0.41 for the within-participants their own memory and keeping all their points (losing 0) each time they correctly remembered special circles, or using reminders, and losing points every time they At this point, we want to clarify that both experiments were conducted remembered (2–8). Contrasting to Experiment 1, this independently from two separate working groups. We follow the same guide- had the advantage that both conditions were equiva- lines of transparency and openness as in Experiment 1; however, some parts lent in terms of the reward participants received, that is of the methodical and statistical approach differ between experiments. Also, note that in the original order Experiment 2 has been performed prior to the outcomes were phrased to appear as either gains or Experiment 1. We changed experiments within the scope of the current paper losses, but objectively they were the same. For example, if as we perceived this order to be more coherent. Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 10 of 18 Fig. 4 Metacognitive Confidence Judgment. Note The metacognitive confidence rating provided us with our metacognitive confidence measure: After participants finished a series of practice trials, they rated their confidence in their ability to accurately remember the delayed intention of dragging the special circles to the respective edges. Participants performed these ratings once (prior to the start of the experiment) separately for the internal and external strategies. For the confidence rating after the forced‑ external trials, participants received the alternate instruction ’Now that you have practiced doing the task using reminders, we would like you to tell us how accurately you can perform the task when you use this strategy’ offered 7 points to use reminders and a participant chose circle, presented in random order). On even-numbered to use reminders and successfully remembered every trials, participants alternated between the forced- special circle, in the gain condition they would earn 70 external and forced-internal trials, with the starting points. Whereas, in an equivalent trial of the loss con- trial type (external or internal) randomized between dition (i.e., 3 points offered to use reminders), they lost participants and counterbalanced between gain/loss 30 points and thus also retained 70 points. By using this conditions. Participants then received experimental scoring scheme, we were expecting to maximize the com- instructions for the other condition (gain or loss). After parability of framing conditions, thus reducing the risk finishing the second experimental block (gain or loss; of potential data noise. Note that this change in scoring 13 trials as above), participants completed two ques- also resulted in Experiment 2 studying the framing effect tionnaires. This was part of the other unrelated project, whereas Experiment 1 investigated the reflection effect addressing a different question which is reported in a (Fagley, 1993). separate paper (see Kirk et al., 2021, for full details). For a demonstration, the entire experiment can be accessed at. Procedure http:// ucl. ac. uk/ sam- gilbe rt/ demos/ CWPK1/ start. html Participants first provided informed consent and then completed six practice trials, with the practice trials following the same procedure as in Experiment 1. How- Reward ever, after each pair of consecutive practice trials with This time we paid participants depending on their task and without reminders, participants made their meta- performance. Paying participants based on their task cognitive judgments reporting how accurately they can performance should ensure that they are more moti- perform the task with the respective strategy. Partici- vated to make optimal choices (see Sachdeva & Gilbert, pants were then randomly assigned to the gain or loss 2020). Implementing this payment allowed us to evaluate condition. In the first experimental block (gain or loss), whether effects of framing observed in Experiment 1 dif - participants performed a total of 13 (7 free, 6 forced) fer in situations where participants have a financial incen - trials. Due to the within-participants implementation tive to choose optimally. To do this, participants were in Experiment 2, participants had to perform trials of told that they were scoring points, where 300 points were both conditions and therefore for practical reasons, equivalent to $1. They received 600 points at the begin - that is to keep the task duration within reasonable ning of the experiment. Then, they were additionally able limits for the participants, we reduced the total trial to earn (or keep) up to 1300 points (i.e., 100 points per number per block. As a result of that, the range of the trial) in each half of the experimental trials. Therefore, indifference points were changed in accordance with the earnings could range between 600 points ($2) and the reduced number of the free-choice trials, that is 3200 points ($10.67). The experiment was advertised as from 1 to 9, in Experiment 1, to 2 to 8, in Experiment having a base payment of $2, which participants received 2. On odd-numbered trials, participants were given a simply for taking part, with the additional earnings sent free choice between using internal memory (10 points to participants afterward as a bonus payment. per special circle) or reminders (2–8 points per special F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 11 of 18 Design and they remembered nearly all of the special circles We entered framing (gain vs. loss) as within-partici- when using reminders in forced-external trials (see pants variable into our design. The variables of interest Fig. 5a). This replicates the accuracy data found in were the same as in Experiment 1, with the indifference Experiment 1 and previous research (Gilbert et al., 2020). points bounded to the range 2 to 8. In addition, measures We submitted the accuracy data to a 2 (condition: forced- regarding the metacognitive judgments were added: internal vs. forced-external; within) × 2 (framing: gain vs. loss; within) ANOVA with repeated measures on both • Internal metacognitive confidence. This is the variables. We again obtained a significant main effect of response made to the metacognitive accuracy predic- condition, F(1, 299) = 991.36, p < 0.001, η = 0.77, but tion following practice trials using internal memory no significant main effect of framing, F(1, 299) = 0.33, (see Fig. 4). p = 0.565, η < 0.01, and a non-significant interaction • External metacognitive confidence. This is the of framing and condition, F(1, 299) = 0.02, p = 0.892, response made to the metacognitive accuracy predic- η < 0.01. This indicates that accuracy was significantly tion following practice trials using reminders. affected by the use of reminders, but not by the framing • Internal metacognitive bias. This is the difference of reminders, just as in Experiment 1. between metacognitive confidence and actual accu - racy on forced-internal trials. A positive number Reminder bias would indicate overconfidence of their own memory Our key hypotheses were tested using the reminder bias abilities. scores. As preregistered, we analyzed the reminder bias • External metacognitive bias. This is the difference scores (OIP minus AIP) with two-tailed one-sample t between metacognitive confidence and actual accu - tests. We observed a significant reminder bias both in the racy on forced-external trials. A positive number gain framing condition, t(299) = 10.32, p < 0.001, d = 0.60, would indicate overconfidence of their performance and in the loss framing condition, t(299) = 4.37, p < 0.001, when using reminders. d = 0.25. To evaluate whether these reminder bias scores differed as a function of framing conditions, we per - formed a preregistered two-tailed paired t test comparing Exclusion criteria reminder bias scores of the gain and loss conditions. This Similar to Experiment 1, participants were excluded if (a) analysis yielded a reminder bias that was, on average, 0.70 their accuracy in the forced-internal condition was lower points greater in the gain condition (M = 1.28, SD = 2.14) than 10%, averaged across the gain and loss conditions; compared to the loss condition (M = 0.58, SD = 2.28), (b) accuracy in the forced-external condition was lower t(299) = 5.05, p < 0.001, d = 0.29. than 70%, averaged across the gain and loss conditions; Given that we manipulated framing within-participants (c) accuracy on the forced-internal trials was higher than in Experiment 2, we obtained two separate reminder bias forced-external trials in either condition; (d) there was a scores for each participant, once for the gain condition negative point biserial correlation between points offered and once for the loss condition. We calculated a prereg- for correct responses on each trial using reminders (2–8) istered Pearson’s product–moment correlation on these and choice of strategy (0 = own memory, 1 = reminders; two scores to check whether individual differences in task this excludes participants who were more likely to set behavior were significantly related across framing condi - reminders when it earned them fewer points, suggest- tions. We found that the reminder bias scores in the two ing random strategy selection); (e) reminder bias score framing conditions were significantly correlated, r = 0.41, (averaged across the gain and loss conditions) exceeded 3 p < 0.001. This indicates that low (high) reminder bias median absolute deviation units (MAD; Leys et al., 2013); scores in one framing condition were associated with low (f ) difference in reminder bias scores between the two (high) scores in the other condition. As in Experiment 1, conditions exceeded 3 MAD units; and (g) internal meta- we exploratorily performed a Pearson’s product–moment cognitive bias score exceeded 3 MAD units. Data collec- correlation between the OIPs and AIPs for both framing tion continued until the study had the appropriate power conditions (see Fig. 5b). There was a significant correla - (N = 300) following exclusion (64 excluded). The data are tion both under gain framing, r = 0.43, p < 0.001, and loss publicly available (https:// osf. io/ 8zvf6/). framing, r = 0.38, p < 0.001, with the difference between correlations being non-significant, Pearson and Filon’s Results Accuracy Please note that Kirk et al. (2021) used the data of the gain condition for Participants remembered almost two-thirds of the spe- their analyses. Therefore, some values concerning offloading behavior in the cial circles using their memory in forced-internal trials, gain condition are identical between the two projects. Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 12 of 18 Fig. 5 Results on Offloading Behavior from Experiment 2. Note a In the left panel, mean accuracy for the forced‑internal and forced‑ external trials in the gain and loss framing condition. Mean optimal (OIPs) and actual indifference points (AIPs) as a function of framing condition in the right panel. The line pattern demonstrates the size of the reminder bias defined as OIP minus AIP. b Correlation between actual and optimal indifference points for the gain and loss framing condition. Error bars represent standard errors (1898) z = 0.83, p = 0.405. (Note that Experiments 1 and and M = − 11.21, SD = 14.97, respectively). This indi - 2 differ in this respect as this time we tested for the dif - cated that participants were significantly underconfident ference between dependent, non-overlapping correla- about their memory abilities and their accuracy using tions using the R package cocor; Diedenhofen & Musch, reminders. Second, we assessed whether participants’ 2015.) Similar to Experiment 1, participants who ben- underconfidence in their internal judgments was associ - efited the most from using reminders (i.e., low OIP) were ated with a higher reminder bias by calculating a Pear- more likely to offload intentions (i.e., low AIP). However, son’s product–moment correlation between the internal in contrast to Experiment 1, this relationship between metacognitive bias score and the reminder bias score actual task performance and offloading behavior did not for both framing conditions (see Fig. 6). A significant differ between framing conditions. Metacognitive bias Note that according to our preregistration we planned to correlate the inter- nal metacognitive bias score with the reminder bias within each condition As reported in the preregistration, we performed three (gain or loss). However, we realized that this could yield spurious self-correla- additional analyses involving metacognitive judgments. tions due to non-independent observations, because the forced-internal accu- First, we performed separate two-tailed one-sample t racy score was used to calculate both the internal metacognitive bias and the reminder bias. To avoid this potential statistical artifact, we deviated from the tests on the internal and external metacognitive bias preregistration by calculating the internal metacognitive bias from the oppo- scores against zero to test whether participants were site experimental condition (i.e., the reminder bias in the gain condition was under- or overconfident using the two strategies. Both correlated with the internal metacognitive bias from the loss condition, and vice versa). This ensured that the correlation analyses met the assumption of the internal, t(299) = − 3.65, p < 0.001, d = 0.21, and independent observations. Results were similar when the analyses were con- external, t(299) = − 12.96, p < 0.001, d = 0.75, metacogni- ducted as specified in the original preregistration. tive bias scores were significant (M = − 6.69, SD = 31.76; F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 13 of 18 Fig. 6 Relationship between Metacognitive Confidence and Offloading Behavior. Note Correlation between individuals’ metacognitive bias and reminder bias scores, separately for both framing conditions. Negative metacognitive bias scores illustrate underconfidence in memory ability and positive reminder bias scores represent overuse of reminders negative correlation was found in both framing condi- p < 0.001. We thus selected the model with the two fixed tions between reminder bias and internal metacognitive factors and compared it with a model including the inter- judgments (gain: r = − 0.23, p < 0.001; loss: r = − 0.19, action of internal metacognitive bias and framing as the p < 0.001), indicating that underconfident participants third fixed effect (AIC: 2603.60), with no significant dif - were more biased toward reminders, regardless of fram- ferences between the models in a likelihood-ratio test, ing condition in free-choice trials. Third, we assessed χ (1) = 0.09, p = 0.761. u Th s, we retained the simpler whether the relationship between reminder bias and model (additive terms only). This suggests that framing internal metacognitive bias differed as a function of adds an additive predictive value to the reminder bias, framing condition by using the R package cocor (test for that is beyond the effect of participants’ metacognitive dependent, non-overlapping correlations; Diedenhofen miscalibration of their internal memory abilities. & Musch, 2015). We found no evidence that framing altered the relationship between internal confidence and Discussion the tendency to use reminders, Fisher’s (1925) z = − 0.62, Consistent with our study’s key hypotheses, Experi- p = 0.537. ment 2 illustrates that while participants choose to use In an exploratory analysis, we further investigated reminders more than would be optimal, this systematic whether framing influences the reminder bias over bias is reduced by the framing of reminders in terms of and above participants’ internal metacognitive bias. To losses. This indicates that, despite receiving financial achieve this, we fitted mixed-effects models (R package incentive to offload optimally, individuals employ a risk- lme4; Bates et al., 2015) on the reminder bias scores with averse approach in situation involving gains but a more random intercepts for the participant effect. To initiate risk-seeking strategy when offloading implies a loss (see the model selection process, we compared the intercept- Kahneman & Tversky, 1979). Furthermore, in the scope only model (Akaike information criterion [AIC]: 2630.50) of participants’ metacognitive confidence ratings, Experi - with a model with internal metacognitive bias as continu- ment 2 demonstrates that irrespective of the framing of ous fixed factor (AIC: 2623.20). We found significant dif - reminders individuals are underconfident about their ferences between the models in a likelihood-ratio test, χ memory abilities and that this erroneous metacognitive (1) = 9.29, p = 0.002, indicating that the reminder bias underconfidence in turn is associated with a greater bias was significantly increased by participants’ undercon - toward using reminders in both framing scenarios, which fidence in their memory abilities We thus selected the replicates the negative relationship between internal model with the one fixed effect and compared it with a metacognitive bias and reminder bias found in Gilbert model which also included the framing (gain vs. loss) as et al. (2020). Further, we show that the framing of remind- another fixed factor (AIC: 2601.70). Again, a likelihood- ers influences participants choice in using reminders over ratio test revealed significant differences, χ (1) = 23.51, and above participants’ internal metacognitive bias. Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 14 of 18 General discussion (Kahneman & Tversky, 1979), as it demonstrates that The present research aimed at achieving more optimal a switch from gain to loss framing also causes a switch decisions in offloading situations, by assessing whether from a risk-averse strategy to a more risk-seeking strat- the framing of reminders (Kahneman & Tversky, 1979) egy. Moreover, our study illustrates that this switch in influences how optimally intentions are offloaded onto strategy also applies to cognitive offloading. external devices. In two experiments, we manipulated However, even though participants were more risk- the framing of reminders in terms of gains or losses, once seeking in situations involving losses, in both experi- between-participants (Experiment 1) and once within- ments the reminder bias was still present in the loss participants (Experiment 2). Experiment 2 further meas- framing condition. Therefore, participants were still ured metacognitive judgments and provided participants risk-averse and preferred to engage in offloading behav - with financial incentives to offload optimally. Participants ior even though reminders implied a loss. That is, despite were less biased toward using reminders when reminders reminders being framed in terms of losses participants were framed as losses instead of gains. This was also true still preferred the outcome that was more certain (using when participants received financial incentives for opti - reminders) over the uncertain and more variable out- mal offloading in Experiment 2. Further, the bias toward come (using internal memory). Hence, we observed a using reminders was higher the more underconfident preference shift but no preference reversal. This suggests participants were about their internal memory abilities, that, in the context of cognitive offloading, the framing of irrespective of the framing of reminders. The framing of choices in terms of losses does not loom large enough to reminders as well as participants’ internal metacogni- reverse the systematic preference for certain outcomes, tive bias were two independent factors, each providing thus leaving some amount of reminder bias. This finding unique predictive value regarding the optimality of par- may be due to domain-related characteristics of offload - ticipants’ offloading choices. ing situations which, by nature, could have impeded the In line with previous literature on decision-making reversal of risk preferences between framing conditions (Kahneman & Tversky, 1979), we observed that in situa- (for a review on moderating variables under gain and tions involving gains, individuals employed a risk-averse loss framing, see Piñon & Gambara, 2005). For example, offloading strategy in situations involving gains and as individuals are usually more sensitive for decisions under a result used more reminders, compared to situations risk, rather than uncertainty (Tversky & Fox, 1995). Thus, where offloading implied a loss. The experiments’ find - adding accuracy-based feedback for each individual, ings are thus consistent to studies demonstrating framing that is ensuring that the probabilities associated with the effects and reflection effects on decision-making in other respective strategies are known, may dissolve, at least, domains (e.g., Bless et al., 1998; Highhouse & Paese, 1996; some domain-specific constraints of offloading situa - Mather et al., 2012; for a review see Baucells & Villaís, tions (which one could argue would then lead to a further 2010; Kühberger, 1998; Piñon & Gambara, 2005; Steiger reduction of the reminder bias). & Kühberger, 2018; Tversky & Kahneman, 1981). How- A second explanation for individuals not being entirely ever, our study is the first to demonstrate effects of gain risk-seeking in the loss framing condition concerns other and loss framing in the context of cognitive offloading. influence factors that were still present in situations A possible explanation for the differences in offload - involving losses, providing a separate effect on the deci - ing strategies between framing conditions concerns sion to use reminders. Specifically, participants’ internal the certainty effect (Kahneman & Tversky, 1979). As metacognitive bias is one such additional factor influ - demonstrated in both experiments, participants in the encing participants’ offloading decisions (Boldt & Gil - gain condition were systematically biased toward using bert, 2019; Engeler & Gilbert, 2020). In Experiment 2, we reminders, thus preferring the option that provided more measured participants’ metacognitive judgments and we certainty regarding the outcome. This indicates that when observed that our participants were underconfident in the choice between reminders or internal memory is their internal memory capabilities and that this internal framed in terms of gains, individuals are risk-averse and metacognitive bias also contributed to the reminder bias, prefer the certain over the risky outcome, which is in line irrespective of the framing of reminders. Interestingly, with the certainty effect (Kahneman & Tversky, 1979). Engeler and Gilbert (2020) indicate that correcting par- In the loss condition, however, the systematic overuse of ticipants’ internal metacognitive judgments via positive reminders was reduced indicating that participants were feedback about their actual task performance may not more willing to take the risky outcome of potentially for- be sufficient to eliminate the reminder bias. That is, even getting a delayed intention and therefore losing a greater though participants no longer needed to be undercon- number of points by using their internal memory. This fident in their memory abilities, they still showed a sys - finding is consistent to the literature on decision-making tematic bias toward using reminders (Engeler & Gilbert, F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 15 of 18 2020). If one combines those findings with the results of Consistent with research on the original optimal our Experiment 2, which suggested that the framing of reminders task (Gilbert et al., 2020), participants were reminders influenced the reminder bias over and above consistently biased toward using reminders in the gain participants’ internal metacognitive bias, one could argue condition of both experiments, thus engaging in offload - that the framing of reminders and participants’ internal ing behavior more than optimal (Scarampi & Gilbert, metacognitive bias are two independent and additive fac- 2020). However, the reminder bias was still evident in tors influencing participants offloading behavior. the loss condition. With this first experimental demon - A third influence factor on the choice of using remind - stration that individuals deviate from optimal decision- ers is cognitive effort. When participants are provided making in situations where offloading involves possible with a performance-based financial incentive, they invest losses, we add to previous literature. Furthermore, as more cognitive effort resulting in an increased overall both experiments varied with respect to multiple factors task performance and a more optimal use of reminders (e.g., different samples; manipulation within-participants (Sachdeva & Gilbert, 2020). Despite providing a perfor- vs. between-participants; different scoring schemes; mance-based financial incentive in our second experi - see above, for further details), we demonstrate that the ment, participants still showed a significant (though systematic overuse of reminders constitutes a funda- reduced) reminder bias when reminders were framed in mental bias in human decision-making that occurs inde- terms of losses. The combination of a performance-based pendently of subtle variations in context (De Martino financial incentive with a loss framing was not sufficient et al., 2006; Hartley & Phelps, 2012; Mulder et al., 2012; to eliminate the reminder bias. This is consistent to the Tom et al., 2007). view that some decisions are rather difficult to debias, especially if relying on a nonlinear relationship between Implications the decision prospects (gain/benefit vs. loss/cost) and Offloading information onto external artifacts can signifi - the subjective values assigned to those prospects (Arkes, cantly help to fulfil future intentions, with participants 1991). Future research should thus continue investigat- consistently remembering more than 90% of inten- ing the joint––and hopefully debiasing––role of framing, tions using reminders in our experiments. This benefit metacognition, and cognitive effort on cognitive offload - of offloading on memory performance replicates results ing, with a specific focus on how to best achieve more found in previous studies (Engeler & Gilbert, 2020; Gil- optimal decisions in offloading situations. bert et al., 2020; Scarampi & Gilbert, 2020). However, Despite the overall reduction of the reminder bias using external devices in daily life does not always incur induced by loss framing, we also observed a rather con- a gain. For example, although offloading can help remem - sistent tendency of overusing reminders across par- ber intentions when memory resources become taxed, it ticipants. In particular, the correlation analysis between takes time and effort to set them up. Even though these individual reminder biases in the gain framing and loss individual costs of setting a reminder may be small, they framing conditions in Experiment 2 indicated that indi- mount to an unacceptable level if reminders were used vidual differences in participants’ offloading strate - for every one of the dozens, maybe hundreds, of things gies were consistent over the two framing conditions. one intends to do in an ordinary day. Other costs of using This may suggest that participants tendency to overuse reminders may only become apparent in the long run. reminders may be partially due to a preference for con- That is, apart from needing more time and effort, individ - sistency in performance, regardless of choice. That is, by uals miss out important cognitive exercise when setting having a general tendency to use reminders, participants up a reminder, possibly affecting cognitive function - are consistent and avoid the higher variability in accu- ing eventually (see Grinschgl et al., 2021b). In fact, the racy that occurs when using an internal memory strat- human brain is highly plastic (Green & Bavelier, 2008), egy. Therefore, participants may be more willing to avoid requiring exercise to establish and maintain proper func- variance in performance even when using their inter- tioning, especially at an older age (Buckner, 2004; Joubert nal memory may result in more correct responses and & Chainay, 2018; Morrison & Chein, 2011; Park & Bis- a higher financial reward. This explanation is consistent chof, 2013; Small, 2001). Therefore, when aiming toward with research proposing that individual’s preferences are an optimal offloading behavior, in real life one faces the risk-averse in terms of mental effort, opting for a fixed challenge how to deal with such costs and benefits of amount of effort rather than a variable amount (Apps using reminders (i.e., how harmful might it be to use the et al., 2015). This preference may be more prevalent when offloading strategy instead of training the brain and how the task demands are difficult, especially when one choice does this relate to the resulting performance). may require significantly more cognitive effort expendi - In our experiments, we demonstrated that while indi- ture, as in the current study. viduals were biased toward using reminders when facing Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 16 of 18 programmed the general framework of the optimal reminders task. MB the prospect of a gain, individuals’ offloading decisions adapted the programming code for Experiment 1. All authors contributed to tended to be more optimal in situations involving losses. the data analysis and were responsible for the interpretation of the results. LF Therefore, framing in the sense of highlighting either drafted the manuscript with the support of all authors. All authors approved the final version of the manuscript for submission. costs or benefits of offloading may also be a promising tool in achieving a more optimal offloading regarding Author information the relation of costs and benefits in daily life. Specifically, Preregistrations as well as data, code, and materials have been made publicly available at OSF; see https:// osf. io/ 8shkf/ (Experiment 1), https:// osf. io/ 8zvf6/ with using framing as a tool one may be able to develop (Experiment 2). Parts of this work were presented at the Cognitive Offloading interventions targeting the perspective individuals adopt Meeting in July 2021 hosted from the University College London and the TeaP in offloading contexts. That is, in situations where peo - 2022 (Conference of Experimental Psychologists) organized by the University of Cologne. This work entails values also presented in a separate project: ple are biased one way or the other (toward or away from https:// doi. org/ 10. 1177/ 17470 21820 970156 offloading), the framing of the instructions could be used to make peoples’ offloading decisions more optimal and Funding Open Access funding enabled and organized by Projekt DEAL. We acknowl‑ this approach could be used to create personalized inter- edge support by Open Access Publishing Fund of University of Tübingen for ventions based on each individual’s particular bias. For covering a part of the article processing charges. example, directing individuals’ focus on the costs that Availability of data and materials each strategy (using reminders or memory) has for their The datasets generated and analyzed during the current study as well as the daily life functioning may encourage them toward more materials are available in the OSF repository, https:// osf. io/ 8shkf/ (Experiment flexible adaptions of their offloading strategy. In a clinical 1), https:// osf. io/ 8zvf6/ (Experiment 2). context, for example, this may be a promising technique in establishing an optimal balance between functional Declarations and compensational rehabilitation for patients with neu- Ethics approval and consent to participate rological disease (see Thöne-Otto & Walther, 2008, for Ethical approval was received from the UCL Research Ethics Committee further details). (1584/003), and participants provided informed consent before participat‑ ing in the study for Experiment 2. Experiment 1 was conducted in accord‑ ance with the APA guidelines for research ethics, and participants provided informed consent before participating. Conclusion With the present research, we argue that the optimality Consent for publication with which individuals engage in offloading behavior can Not applicable. be shaped by the way offloading situations are framed. Competing interests When situations are framed in terms of gains, reminders The authors declare that they have no competing interests. are preferred, and when situations are framed in terms of Author details losses, individuals are less risk-averse and use reminders Department of Psychology, University of Tübingen, Tübingen, Germany. more optimally. Whereas previous attempts in achieving 2 Institute of Cognitive Neuroscience, University College London, London, UK. more optimal decisions in offloading situations have been Received: 21 October 2021 Accepted: 20 June 2022 following a mainly metacognitive approach (Engeler & Gilbert, 2020; Gilbert et al., 2020; Risko & Gilbert, 2016), the present study adds the concept of framing to this research area. We demonstrate that individuals’ offload - References ing strategies under gain and loss framing are affected Apps, M. A. J., Grima, L. L., Manohar, S., & Husain, M. (2015). The role of cogni‑ in line with the risk preferences predicted from previ- tive effort in subjective reward devaluation and risky decision‑making. Scientific Reports. https:// doi. org/ 10. 1038/ srep1 6880 ous literature on decision-making (Kahneman & Tversky, Arkes, H. R. (1991). Costs and benefits of judgment errors: Implications for 1979), leading to more optimal decisions when empha- debiasing. Psychological Bulletin, 110(3), 486–498. https:// doi. org/ 10. 1037/ sizing the costs (losses) rather than the benefits (gains) 0033‑ 2909. 110.3. 486 Ballard, D. H., Hayhoe, M. M., & Pelz, J. B. (1995). Memory representations in of offloading behavior. Thus, we claim that it is essential natural tasks. Journal of Cognitive Neuroscience, 7(1), 66–80. to move forward toward an inclusive theory of offload - Bates, D., Maechler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed‑ effects ing behavior, to expand our knowledge on how to best models using lme4. Journal of Statistical Software, 67(1), 1–48. Baucells, M., & Villasís, A. (2010). Stability of risk preferences and the reflection improve humans’ decision-making in offloading contexts. effect of prospect theory. Theory and Decision, 68(1–2), 193–211. https:// doi. org/ 10. 1007/ s11238‑ 009‑ 9153‑3 Acknowledgements Bless, H., Betsch, T., & Franzen, A. (1998). Framing the framing effect: The impact Not applicable. of context cues on solutions to the ‘Asian disease’ problem. European Journal of Social Psychology, 28(2), 287–291. https:// doi. org/ 10. 1002/ (SICI) Author contributions 1099‑ 0992(199803/ 04) 28:2% 3c287:: AID‑ EJSP8 61% 3e3.0. CO;2‑U LF, AF, MB, and FP developed the study concept, contributed to the study Boag, R. J., Strickland, L., Heathcote, A., Neal, A., & Loft, S. (2019). Cognitive design, and were responsible for testing and data collection of Experiment 1. control and capacity for prospective memory in complex dynamic CW and SG developed the study concept, contributed to the study design, and were responsible for testing and data collection of Experiment 2. SG F röscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 17 of 18 environments. Journal of Experimental Psychology: General, 148(12), Highhouse, S., & Paese, P. W. (1996). Problem domain and prospect frame: 2181–2206. https:// doi. org/ 10. 1037/ xge00 00599 Choice under opportunity versus threat. Personality and Social Psychology Boldt, A., & Gilbert, S. J. (2019). Confidence guides spontaneous cognitive Bulletin, 22(2), 124–132. https:// doi. org/ 10. 1177/ 01461 67296 222002 offloading. Cognitive Research: Principles and Implications, 4(1), 45. https:// Joubert, C., & Chainay, H. (2018). Aging brain: The effect of combined cognitive doi. org/ 10. 1186/ s41235‑ 019‑ 0195‑y and physical training on cognition as compared to cognitive and physical Buckner, R. L. (2004). Memory and executive function in aging and AD: training alone—a systematic review. Clinical Interventions in Aging, 13, Multiple factors that cause decline and reserve factors that compensate. 1267–1301. https:// doi. org/ 10. 2147/ CIA. S1653 99 Neuron, 44(1), 195–208. https:// doi. org/ 10. 1016/j. neuron. 2004. 09. 006 Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision Cherry, K. E., & LeCompte, D. C. (1999). Age and individual differences in under risk. Econometrica, 47(2), 263–291. https:// doi. org/ 10. 2307/ 19141 85 prospective memory. Psychology and Aging, 14, 60–76. https:// doi. org/ 10. Kirk, P. A., Robinson, O. J., & Gilbert, S. J. (2021). Trait anxiety does not correlate 1037/ 0882‑ 7974. 14.1. 60 with metacognitive confidence or reminder usage in a delayed inten‑ Crump, M. J. C., McDonnell, J. C., & Gureckis, T. M. (2013). Evaluating Amazon’s tions task. Quarterly Journal of Experimental Psychology, 74(4), 634–644. mechanical turk as a tool for experimental behavioral research. PLoS ONE, https:// doi. org/ 10. 1177/ 17470 21820 970156 8(3), e57410. https:// doi. org/ 10. 1371/ journ al. pone. 00574 10 Kliegel, M., McDaniel, M. A., & Einstein, G. O. (2000). Plan formation, retention, De Martino, B., Kumaran, D., Seymour, B., & Dolan, R. J. (2006). Frames, biases and execution in prospective memory: A new approach and age‑related and rational decision‑making in the human brain. Science, 313(5787), effects. Memory and Cognition, 28(6), 1041–1049. https:// doi. org/ 10. 3758/ 684–687. https:// doi. org/ 10. 1126/ scien ce. 11283 56BF032 09352 Diedenhofen, B., & Musch, J. (2015). Cocor: A comprehensive solution for the Kool, W., McGuire, J. T., Rosen, Z. B., & Botvinick, M. M. (2010). Decision making statistical comparison of correlations. PLoS ONE, 10(4), e0121945. https:// and the avoidance of cognitive demand. Journal of Experimental Psychol- doi. org/ 10. 1371/ journ al. pone. 01219 45 ogy: General, 139, 665–682. https:// doi. org/ 10. 1037/ a0020 198 Einstein, G. O., & McDaniel, M. A. (1990). Normal aging and prospective Kühberger, A. (1998). The influence of framing on risky decisions: A Meta‑anal‑ memory. Journal of Experimental Psychology: Learning, Memory, and Cogni- ysis. Organizational Behavior and Human Decision Processes, 75(1), 23–55. tion, 16(4), 717–726. https:// doi. org/ 10. 1037/ 0278‑ 7393. 16.4. 717https:// doi. org/ 10. 1006/ obhd. 1998. 2781 Ellis, J., Kvavilashvili, L., & Milne, A. (1999). Experimental tests of prospective Levin, I. P., Gaeth, G. J., Schreiber, J., & Lauriola, M. (2002). A new look at framing remembering: The influence of cue ‑ event frequency on performance. effects: Distribution of effect sizes, individual differences, and independ‑ British Journal of Psychology, 90(1), 9–23. https:// doi. org/ 10. 1348/ 00071 ence of types of effects. Organizational Behavior and Human Decision 26991 61233 Processes, 88(1), 411–429. https:// doi. org/ 10. 1006/ obhd. 2001. 2983 Engeler, N. C., & Gilbert, S. J. (2020). The effect of metacognitive training on Leys, C., Ley, C., Klein, O., Bernard, P., & Licata, L. (2013). Detecting outliers: Do confidence and strategic reminder setting. PLoS ONE, 15(10), e0240858. not use standard deviation around the mean, use absolute deviation https:// doi. org/ 10. 1371/ journ al. pone. 02408 58 around the median. Journal of Experimental Social Psychology, 49(4), Fagley, N. S. (1993). A note concerning reflection effects versus framing effects. 764–766. https:// doi. org/ 10. 1016/j. jesp. 2013. 03. 013 Psychological Bulletin, 113(3), 451–452. https:// doi. org/ 10. 1037/ 0033‑ 2909. Linares, D., & López‑Moliner, J. (2019). Quickpsy: Fits psychometric functions for 113.3. 451 multiple groups (version 0.1.5.1). [R package]. CRAN. https:// dlina res. org/ Faul, F., Erdfelder, E., Lang, A.‑ G., & Buchner, A. (2007). G*Power 3: A flexible quick psy. html. statistical power analysis program for the social, behavioral, and biomedi‑ Mather, M., Mazar, N., Gorlick, M. A., Lighthall, N. R., Burgeno, J., Schoeke, A., & cal sciences. Behavior Research Methods, 39(2), 175–191. https:// doi. org/ Ariely, D. (2012). Risk preferences and aging: The “certainty effect” in older 10. 3758/ BF031 93146 adults decision making. Psychology and Aging, 27(4), 801–816. https:// doi. Fisher, R. A. (1925). Statistical methods for research workers. Oliver and Boyd. org/ 10. 1037/ a0030 174 http:// psych class ics. yorku. ca/. Meyerhoff, H. S., Grinschgl, S., Papenmeier, F., & Gilbert, S. J. (2021). Individual Gilbert, S. J. (2015a). Strategic offloading of delayed intentions into the exter ‑ differences in cognitive offloading: A comparison of intention offloading, nal environment. The Quarterly Journal of Experimental Psychology, 68(5), pattern copy, and short‑term memory capacity. Cognitive Research: Princi- 971–992. https:// doi. org/ 10. 1080/ 17470 218. 2014. 972963 ples and Implications. https:// doi. org/ 10. 1186/ s41235‑ 021‑ 00298‑x Gilbert, S. J. (2015b). Strategic use of reminders: Influence of both domain‑ Morrison, A. B., & Chein, J. M. (2011). Does working memory training work? general and task‑specific metacognitive confidence, independent of The promise and challenges of enhancing cognition by training working objective memory ability. Consciousness and Cognition, 33, 245–260. memory. Psychonomic Bulletin & Review, 18(1), 46–60. https:// doi. org/ 10. https:// doi. org/ 10. 1016/j. concog. 2015. 01. 0063758/ s13423‑ 010‑ 0034‑0 Gilbert, S. J., Bird, A., Carpenter, J. M., Fleming, S. M., Sachdeva, C., & Tsai, P.‑ C. Mulder, M. J., Wagenmakers, E. J., Ratcliff, R., Boekel, W., & Forstmann, B. U. (2020). Optimal use of reminders: Metacognition, effort, and cognitive (2012). Bias in the brain: A diffusion model analysis of prior probability offloading. Journal of Experimental Psychology: General, 149, 501–517. and potential payoff. The Journal of Neuroscience, 32(7), 2335–2343. https:// doi. org/ 10. 1037/ xge00 00652https:// doi. org/ 10. 1523/ JNEUR OSCI. 4156‑ 11. 2012 Gray, W. D., Sims, C. R., Fu, W.‑ T., & Schoelles, M. J. (2006). The soft constraints Papenmeier, F. (2018). Powerbydesign: Power estimates for ANOVA designs (ver‑ hypothesis: A rational analysis approach to resource allocation for interac‑ sion 1.0.4) [R package]. CRAN. https:// cran.r‑ proje ct. org/ packa ge= power tive behavior. Psychological Review, 113(3), 461–482. https:// doi. org/ 10. bydes ign. 1037/ 0033‑ 295X. 113.3. 461 Park, D. C., & Bischof, G. N. (2013). The aging mind: Neuroplasticity in response Green, C. S., & Bavelier, D. (2008). Exercising your brain: A review of human to cognitive training. Dialogues in Clinical Neuroscience, 15(1), 109–119. brain plasticity and training‑induced learning. Psychology and Aging, Pearson, K., & Filon, L. N. G. (1898). Mathematical contributions to theory of 23(4), 692–701. https:// doi. org/ 10. 1037/ a0014 345 evolution: IV. On the probable errors of frequency constants and on the Grinschgl, S., Meyerhoff, H. S., & Papenmeier, F. (2020). Interface and interaction influence of random selection and correlation. Philosophical Transactions design: How mobile touch devices foster cognitive offloading. Comput- of the Royal Society of London, Series A, 191, 229–311. https:// doi. org/ 10. ers in Human Behavior, 108, 106317. https:// doi. org/ 10. 1016/j. chb. 2020. 1098/ rsta. 1898. 0007 106317 Piñon, A., & Gambara, H. (2005). A meta‑analytic review of framing effect: Risky, Grinschgl, S., Meyerhoff, H. S., Schwan, S., & Papenmeier, F. (2021a). From meta‑ attribute and goal framing. Psicothema, 17(2), 325–331. cognitive beliefs to strategy selection: Does fake performance feedback Risko, E. F., & Dunn, T. L. (2015). Storing information in‑the ‑ world: Metacogni‑ influence cognitive offloading. Psychological Research Psychologische tion and cognitive offloading in a short ‑term memory task. Consciousness Forschung. https:// doi. org/ 10. 1007/ s00426‑ 020‑ 01435‑9 and Cognition, 36, 61–74. https:// doi. org/ 10. 1016/j. concog. 2015. 05. 014 Grinschgl, S., Papenmeier, F., & Meyerhoff, H. S. (2021b). Consequences of Risko, E. F., & Gilbert, S. J. (2016). Cognitive offloading. Trends in Cognitive Sci- cognitive offloading: Boosting performance but diminishing memory. ences, 20(9), 676–688. https:// doi. org/ 10. 1016/j. tics. 2016. 07. 002 Quarterly Journal of Experimental Psychology, 74(9), 1477–1496. https:// doi. Sachdeva, C., & Gilbert, S. J. (2020). Excessive use of reminders: Metacognition org/ 10. 1177/ 17470 21821 10080 60 and effort ‑minimisation in cognitive offloading. Consciousness and Cogni- Hartley, C. A., & Phelps, E. A. (2012). Anxiety and decision‑making. Biological tion, 85, 103024. https:// doi. org/ 10. 1016/j. concog. 2020. 103024 Psychiatry, 72(2), 113–118. https:// doi. org/ 10. 1016/j. biops ych. 2011. 12. 027 Fröscher et al. Cognitive Research: Principles and Implications (2022) 7:61 Page 18 of 18 Scarampi, C., & Gilbert, S. J. (2020). The effect of recent reminder setting on subsequent strategy and performance in a prospective memory task. Memory, 28(5), 677–691. https:// doi. org/ 10. 1080/ 09658 211. 2020. 17649 74 Small, S. A. (2001). Age‑related memory decline: Current concepts and future directions. Archives of Neurology, 58(3), 360–364. https:// doi. org/ 10. 1001/ archn eur. 58.3. 360 Steiger, A., & Kühberger, A. (2018). A meta‑analytic re ‑appraisal of the framing effect. Zeitschrift Für Psychologie, 226(1), 45–55. https:// doi. org/ 10. 1027/ 2151‑ 2604/ a0003 21 Thöne‑ Otto, A. I. T., & Walther, K. (2008). Assessment and treatment of prospec‑ tive memory disorders in clinical practice. In M. Kliegel, M. A. McDaniel, & G. O. Einstein (Eds.), Prospective Memory: Cognitive, Neuroscience, Devel- opmental, and Applied Perspectives (pp. 321–345). Taylor & Francis Group/ Lawrence Erlbaum Associates. Tom, S. M., Fox, C. R., Trepel, C., & Poldrack, R. A. (2007). The neural basis of loss aversion in decision‑making under risk. Science, 315(5811), 515–518. https:// doi. org/ 10. 1126/ scien ce. 11342 39 Tversky, A., & Fox, C. R. (1995). Weighing risk and uncertainty. Psychological Review, 102(2), 269–283. https:// doi. org/ 10. 1037/ 0033‑ 295X. 102.2. 269 Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychol‑ ogy of choice. Science, 211(4481), 453–458. https:// doi. org/ 10. 1126/ scien ce. 74556 83 New York. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub‑ lished maps and institutional affiliations.

Journal

Cognitive Research:Principles and Implications – Springer Journals

Published: Jul 16, 2022

Keywords: Framing; Offloading; Reminders; Metacognition; Prospective memory

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Framing cognitive offloading in terms of gains or losses: achieving a more optimal use of reminders

Framing cognitive offloading in terms of gains or losses: achieving a more optimal use of reminders

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Framing cognitive offloading in terms of gains or losses: achieving a more optimal use of reminders

Framing cognitive offloading in terms of gains or losses: achieving a more optimal use of reminders

References (52)

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies