Access the full text.
Sign up today, get DeepDyve free for 14 days.
Adults are faster and more accurate at detecting changes to animate compared to inanimate stimuli in a change-detection paradigm. We tested whether 11-month-old children detected changes to animate objects in an image more reliably than they detected changes to inanimate objects. During each trial, infants were habituated to an image of a natural scene. Once the infant habituated, the scene was replaced by a scene that was identical except that a target object was removed. Infants dishabituated significantly more often if an animate target had been removed from the scene. Dishabituation results suggested that infants, like adults, preferentially attend to animate rather than to inanimate objects. Keywords animacy, change-detection, change-blindness, social attention Date received: May 04, 2021; Accepted: June 09, 2021 Humans are efficient at detecting animate agents and attend stimuli compared to inanimate objects and use the same corti- to them preferentially (New et al., 2007; Ro et al., 2007). cal routes to process face stimuli as adults would (Buiatti et al., Animacy detection is the ability to quickly distinguish between 2019; Fantz, 1963)). Even when stimulus familiarity is con- what is animate and what is inanimate. Animacy detection was trolled for, 9-week-old infants smile and vocalize more toward functional and allowed for fast identification of social partners a person than a doll, indicating that they recognize which one is and potential predators and prey in the environment of evolu- a social partner (Legerstee et al., 1987). By 12 weeks of age, tionary adaptedness. It is assumed that selection pressures have infants look longer at a person than at a toy monkey (Brazelton facilitated the evolution of both efficient detection of animate et al., 1974). objects and selective attention to animate stimuli. This atten- This sensitivity to animacy is not only based on appearance; tion is recruited spontaneously, regardless of the context and motion alone also acts as a cue to animacy. Newborn infants current goals of the observer (New et al., 2007), and is irresis- can already distinguish between animate and inanimate motion tible (Scholl & Gao, 2013). Selective attention to animacy is (Di Giorgio et al., 2017, 2021) and infants show a looking time not exclusive to humans but apparent across species. Newborn preference for animate motion (Frankenhuis et al., 2013; chicks, like humans, show an attunement to animacy (Regolin Rochat et al., 1997). Likewise, newborn chicks preferentially et al., 2000; Salva et al., 2015; Vallortigara, 2012). Failure to attend to animate agents versus inanimate motion (Lorenzi detect a potential predator could have resulted in injury or et al., 2017; Rosa-Salva et al., 2018). Infants are sensitive to death, providing a selection pressure for the detection of ani- mate stimuli (Ohman et al., 2001). Attention to humans, of course, was particularly relevant as they were potential friends, 1 Department of Psychology, Neuroscience and Behaviour, McMaster foes, or mating opportunities. University. Hamilton, Ontario, Canada Corresponding Author: Ruth Hofrichter, Department of Psychology, Neuroscience and Behaviour, Early Attention to Animacy McMaster University, 1280 Main Street West, Hamilton, Ontario, Canada Preferential attention to humans and other animate objects L8S 4K1. develops early. Newborn infants prefer looking at face-like Email: email@example.com Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https://creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 Evolutionary Psychology animacy cues such as self-propulsion (Di Giorgio et al., 2017; colleagues (2007). Young infants have been shown to be sen- Mascalzoni et al., 2010), expect animate agents to act rationally sitive to animate motion (Frankenhuis et al., 2013; Rochat (Csibra et al., 1999), in a goal-directed manner (Wagner & et al., 1997) but no studies have tested infants’ preferential Carey, 2005), and to adhere to the laws of physics (Arterberry attention to animate objects by using infants’ ability to detect & Bornstein, 2002; Spelke, Phillips, & Woodward, 1995). change in static images. To adapt the change-detection para- Seven-month-old infants are surprised if objects move on their digm for use with infants, we used a habituation paradigm. We own but not when people do (Tra ¨uble et al., 2014; Woodward included 11-month-olds as our participants because infants this et al., 1993). This indicates that within the first year of life, age have been shown to be attuned to animacy (Rochat et al., infants are not only interested in animate stimuli but develop a 1997) and are able to sustain attention longer than younger more sophisticated understanding of what stimuli have agency infants. and which do not. The stimuli in our study were a subset of the scenes used by When it comes to imitation, Meltzoff reported that New et al. (2007). Participants were habituated to Scene A, 18-month-olds were six times more likely to complete then presented with Scene A’. In our study there was always actions modelled by a human than those modelled by a robot a change between Scene A and A’. Using an eye tracker, we (Meltzoff, 1995). Legerstee and colleagues replicated this measured whether the child dishabituated to the change, sug- finding with 10-month-old infants (Legerstee & Markova, gesting that the child had noticed a difference between Scene A 2008). Together, these studies indicate that preverbal infants and A’. We predicted that infants, like adults, would be better at have already developed a representation of animate beings detecting changes to animate compared to inanimate entities, that combines static attributes with dynamic motion. Looking suggesting heightened attention. time preferences for animate stimuli suggest that infants pay particular attention to humans and other animate agents. In investigations with adults, a change-detection paradigm offers Method even stronger evidence of preferential attention for animate Data collection complied with current APA Ethical Principles stimuli. of Psychologists and Code of Conduct, and all measures, manipulations and exclusions in the study are disclosed. Measuring Attention to Animacy Using a Change-Detection Paradigm Participants New and colleagues (2007) used a change-detection paradigm Thirty-six 11-month-old infants (20 female; M ¼ 11.13 to test whether changes to animate stimuli would be detected months, SD ¼ .36) were recruited through an existing database faster and more frequently than changes to inanimate stimuli. for child participants. Parents reported their children’s ethnicity Adult participants were presented with an image of a complex as Hispanic (two), Asian (four) or Caucasian (30). Parents and naturalistic scene (Scene A) for 250 ms which was then masked their children were compensated for their time with $10. Data by a blank screen. Then a second image was presented (Scene of 10 participants was excluded from analyses due to computer/ A’). Scene A’ was either identical to Scene A, or one object had calibration failure (three) children not completing trials (three) been removed from the scene. The object that was removed or parents pointing out objects in the scenes (four). A sensitiv- was either animate or inanimate. Scene A and A’ alternated on ity power analysis indicated that the sample (N ¼ 26) had 80% a loop until participants could identify the change. Results power to detect one-tailed, within-subject differences of indicated that adults were faster and more likely to detect ani- d ¼ 0.5 or higher. The sensitivity power analysis was con- mate compared to inanimate changes. Participants were more ducted using the pwr package in R 4.0.0. likely to experience change-blindness, failing to detect changes, when the changed object was inanimate. This superior Stimuli detection of changes in animate stimuli was independent of The stimuli were color images of complex, natural scenes (see expertise. New and colleagues (2007) used vehicles as one of Figure 1 for examples; all stimuli are shown in Appendix A). the stimulus categories, reasoning that if expertise was driving We selected eight pairs (Scene A & A’) from New et al. (2007) the effect, they should see an advantage for detecting changes original stimuli set: four pairs associated with changes to ani- to vehicles. Results showed no such advantage, despite the fact mate entities, and four scenes associated with changes to inan- that in our current environment, vehicles pose a potential imate entities. Animate entities that were removed were a threat. Participants were not simply attending to entities that horse, a lion, a man, and an officer. Inanimate entities were a moved or could be potentially dangerous. TV, a tree, a cup, and a building (for example in Figure 1A, the lion disappears in Scene A’. In Figure 1B, the tree disappears). Current Study Areas of interest (AOI) were created around the target (see The current study was designed to test for preferential attention Figure 1). We chose images in which the changing element to animate over inanimate objects in 11-month-olds using a was large so that infants would be more likely to notice the change-detection paradigm akin to that used by New and change. The average size of AOIs for animate and inanimate Hofrichter et al. 3 Figure 1. Examples of stimuli with areas of interest highlighted. entities was equal (average area of AOI ¼ 0.74 inches). Fur- ther, for the inanimate change images, we selected scenes that did not also include any non-changing animate objects because they could potentially distract participants from noticing the inanimate change (Altman et al., 2016). We defined Habituation as three trials with averaged looking time of less than 50% of the average looking time of the first 3 trials. We also required children to look at the screen for at least three cycles of Scene A before attention dropped to meet habituation criteria. An experimenter who stood beside the infant throughout the experiment noted how Figure 2. Trial structure. many cycles of Scene A the infant attended to and advanced the trial to Scene A’ once the participant had habituated. The eye tracker captured whether the child looked at AOIs. We At the beginning of each trial, an attention-grabber only included trials in our analysis if infants had looked at the appeared: a video clip of a duck shaped rattle that shakes and area of interest in Scene A at least once, to ensure that infants makes a loud, quacking sound. The attention-grabber used was had an opportunity to notice the target in Scene A before it one included with the Tobii studio software and was shown was removed in Scene A’. continuously until it successfully directed the child’s attention Dishabituation was noted only if 1) the child had met the towards the screen. As soon as the child was looking directly at habituation criterion, 2) looking time recovered to above 50% the screen, Scene A was displayed. of baseline looking time at test and, 3) the child looked at the Scene A was displayed for 15 seconds. After 15 seconds, a area of interest in Scene A’, the location of the removed object, blank screen masked the screen for 250 ms, then Scene A at least once. returned. This cycle continued until the infant had habituated to Scene A. Scene A was shown a minimum of six times and a maximum of 12 times. Our inclusion criterion was that if a Procedure child did not habituate within 12 cycles, the trial would be excluded from analysis. Once habituation was achieved, Scene At the beginning of the session, parents completed a demo- A was masked again with a blank screen. Lastly, Scene A’ was graphic survey providing information on the age, sex, and eth- presented for 15 seconds, then the trial was over (see Figure 2). nicity of their infant. To track infants’ eye movements, we used a Tobii T60XL eye tracker (24-inch screen; 1920 1080 pixels widescreen monitor). The infant was positioned in front of the Results eye tracker on the parent’s lap approximately 24 inches away from the screen. Parents wore sunglasses to ensure that their Participants were assigned a score of 0 or 1 for each trial. A eyes were not detected by the eye tracker. score of 1 indicated that they dishabituated at test, while a score 4 Evolutionary Psychology Figure 3. Dishabituation rates for animate versus inanimate changes. Figure 4. Looking time toward animate versus inanimate areas of Error bars represent 95% confidence intervals. interest (AOI). Error bars represent 95% confidence intervals. See Figure 4). Children spent slightly more time looking at of 0 indicated no dishabituation. Every child could receive a Animate stimuli (M ¼ 225.48, SD ¼ 511.35) than Inanimate maximum of eight scores (four animate trials, four inanimate stimuli (M ¼ 106.81, SD ¼ 323.13). trials). On average, 11-month-olds completed 3.4 (out of 4) animate (SD ¼ 1.12) and 2.04 (out of 4) inanimate (SD ¼ 1.74) trials. Further, on average, participants watched 6.34 cycles Discussion of animate scenes (SD ¼ .55) and 6.23 of inanimate scenes (SD ¼ .51) until habituation was achieved. Our results revealed higher dishabituation rates for animate change trials compared to inanimate change trials. Infants were more likely to notice changes if an animate object was removed Analytic Strategy from the scene compared to an inanimate object, which is We conducted a mixed-effects linear model to analyze chil- consistent with the findings reported by New and colleagues dren’s dishabituation rates to animate versus inanimate stimuli. (2007). These results suggest that in the first year of life, infants The model included a fixed effect for Category (Animate vs. already have the attentional preference for animate objects that Inanimate trials) and a random intercept for Subject. We used a adults do. linear model instead of a logistic model as some have argued New and colleagues (2007) argued that an attentional pre- that linear models may be optimal for analyzing binary ference for animate objects had functional significance in our responses (Gomila, 2019). Additionally, linear models are less evolutionary past and could not be explained by familiarity. hampered by smaller sample sizes. Animate objects in our study included a horse and a lion, ani- We conducted a second mixed-effects linear model to ana- mals that infants in our study might have never encountered lyze children’s looking time toward animate versus inanimate before. Inanimate objects included objects like a TV or a cup. AOIs. This model also included a fixed effect for Category Since most infants in our study are likely to have seen and (Animate vs. Inanimate) and a random intercept for Subject. interacted with a TV or a cup, familiarity cannot account for All analyses were conducted using the lmer command in the our results. Rather, preferences for detecting animate changes lme4 package in R 4.0.0. The default optimizer was used. early in life add further evidence to New and colleagues’ (2007) theory that preferential attention for animacy is evolu- tionarily driven. This is in line with evidence that other species, Findings such as chicks, show a preference for animate stimuli from There was a significant effect of Category on Dishabituation birth onward (Lorenzi et al., 2017; Regolin et al., 2000; rates (Wald w (1) ¼ 4.83, p ¼ 0.03, Hedges’ g ¼ 0.36, 95% CI Rosa-Salva et al., 2018; Salva et al., 2015; Vallortigara, 2012). [0.04, 0.68]; See Figure 3). Children dishabituated more often This is the only experiment that we know of that has adapted to Animate stimuli (M ¼ 0.28, SD ¼ 0.45) than Inanimate the adult change-detection paradigm for use with infants. That stimuli (M ¼ 0.14, SD ¼ 0.35). However, there was only a said, Wang and Baillargeon (2006) used a live puppet show marginal effect of Category on Looking Time (Wald w (1) ¼ style display with occlusion and covering events to show that 2.87, p ¼ 0.09, Hedges’ g ¼ 0.27, 95% CI [0.05, 0.59]; 11-month-olds can detect a change in the height of an object Hofrichter et al. 5 during an occlusion but not during a covering event. The video and at test, version A’ would be shown with an animate authors concluded that infants succeeded in detecting a change or inanimate entity removed. If infants dishabituated at test, it to an object’s height only when watching an event in which the would indicate that they noticed the change. infant perceives height to be a relevant variable. They failed to Further, while we ensured that AOIs were equal in size detect the same change in an event where height had not yet between the animate and inanimate category, we did not con- been identified by the infant as a relevant variable. Similarly, trol for the location of the changing objects. Two of our inan- we found a greater frequency for change detection when the imate AOIs (building and cup) were located at the forefront of object had functional significance to the infant (an animate the scene. Our animate AOIs tended to be located closer to the object) than when the object did not (an inanimate object). back of the scene. Additionally, half of our animate entities We did not find significant differences in looking time to were shown in mid-motion. The bias for detecting animate animate versus inanimate objects. Based on previous research, changes could be a result of perceiving these entities in motion. we did not have an a priori expectation of looking time differ- This could present a potential confound and should be con- ing between animate and inanimate entities. However, previous trolled for in future studies. research on infants’ looking time towards animate versus inan- Another possible confound in the current study was low- imate moving objects has shown conflicting findings. Newborn level differences between images of scenes, such as variance infants have been shown to spend more time looking at static in luminance and complexity. New and colleagues (2007) images of schematic faces versus random patterns (Fantz, tested whether such low-level visual characteristics predicted 1963), and infants can distinguish animate from inanimate performance. They did not find any significant effects. How- objects and even bind animacy motion cues to animals (Tra ¨uble ever, although we used the same stimuli as New and col- & Pauen, 2011). While Frankenhuis and colleagues (2013) leagues, we did not control for these low-level characteristics found that 4- and 10-month-old infants strongly preferred look- in our analysis, so we cannot say for certain whether they ing at animate motion, Rochat and colleagues (1997) only presented a confound in our study. It is possible that adults and found this pattern for 3-month-old infants and reported a switch infants are affected differently by low-level features. An in preferential attention towards animate motion around 5–6 improved follow-up experiment might employ a between- months of age. Rochat and colleagues argued that by 6 months subject design using the same scenes in both groups with an of age, infants readily understood the social contingencies animate object removed in one group and an inanimate object underlying the animate display and therefore spent more time in the other. One of our aims with the current study was to looking at the inanimate display, as if scrutinizing the display replicate and expand on the study by New and colleagues for some meaning or pattern. Since infants in our study were (2007), so we used the same stimuli that they had used. How- shown Scene A for a minimum of 45 seconds during habitua- ever, a study that used the same scenes in the animate and tion, they had ample time to explore all objects in the image. inanimate trials would hold a number of potentially relevant Even if infants were initially drawn to animate AOIs, after variables equal across trial types, such as location, lighting, identifying and processing animate objects, they could have subject matter, and implied motion. then moved on to visually explore other objects. If this were the case, we would not see significant differences in looking Conclusions time across categories of AOIs. Our results show evidence for an early-developing attentional preference for animate objects. By the age of 11 months, Limitations and Future Directions infants dishabituate significantly more often for changes to Many participants became bored easily and some did not com- animate compared to inanimate objects. This finding agrees plete all eight trials of our study. Using inattentional blindness with adults’ faster and better detection rates for animate videos akin to those used by Simons and Chabris (1999), changes in New and colleagues’ (2007) change-blindness para- instead of static images might be more interesting to infants. digm. Infants, like adults, seem to prioritize attention to ani- In a future study, children could be habituated to version A of a mate entities and this result cannot be explained by familiarity. 6 Evolutionary Psychology Appendix Figure A1. Overview of stimuli set used in this study. Hofrichter et al. 7 Figure A1. (Continued). 8 Evolutionary Psychology Gomila,R.(2019,July11). Logistic or linear? Estimating causal Declaration of Conflicting Interests effects of treatments on binary outcomes using regression analysis The author(s) declared no potential conflicts of interest with respect to [Preprint]. https://doi.org/10.31234/osf.io/4gmbv the research, authorship, and/or publication of this article. Legerstee, M., & Markova, G. (2008). Variations in 10-month-old infant imitation of people and things. Infant Behavior and Devel- Funding opment, 31(1), 81–91. The author(s) disclosed receipt of the following financial support for Legerstee, M., Pomerleau, A., Malcuit, G., & Feider, H. (1987). The the research, authorship, and/or publication of this article: This development of infants’ responses to people and a doll: Implica- research was funded by a Natural Sciences and Engineering Research Council (Canada) grant to MDR. tions for research in communication. Infant Behavior and Devel- opment, 10(1), 81–95. Lorenzi, E., Mayer, U., Rosa-Salva, O., & Vallortigara, G. (2017). ORCID iDs Dynamic features of animate motion activate septal and preoptic Ruth Hofrichter https://orcid.org/0000-0002-3301-4113 areas in visually naı ¨ve chicks (Gallus gallus). Neuroscience, 354, Marcus N. Morrisey https://orcid.org/0000-0002-6998-957X M. D. Rutherford https://orcid.org/0000-0002-7685-8076 54–68. Mascalzoni, E., Regolin, L., & Vallortigara, G. (2010). Innate sensi- tivity for self-propelled causal agency in newly hatched chicks. Notes Proceedings of the National Academy of Sciences, 107(9), 4483–4485. Note Meltzoff, A. N. (1995). Understanding the intentions of others: re- 1. Originally, both models were fitted with a random intercept for enactment of intended acts by 18-month-old children. Develop- Stimulus to capture adjustments in the intercept for each scene. mental psychology, 31(5), 838. However, the random intercept contributed non-significant var- New, J., Cosmides, L., & Tooby, J. (2007). Category-specific attention iance in both models (ps > 0.15) so was removed for the final for animals reflects ancestral priorities, not expertise. Proceedings analysis. of the National Academy of Sciences, 104(42), 16598–16603. Ohman, A., Flykt, A., & Esteves, F. (2001). Emotion drives attention: References Detecting the snake in the grass. Journal of Experimental Psychol- Altman, M. N., Khislavsky, A. L., Coverdale, M. E., & Gilger, J. W. ogy: General, 130(3), 466. (2016). Adaptive attention: How preference for animacy impacts Regolin, L., Tommasi, L., & Vallortigara, G. (2000). Visual percep- change detection. Evolution and Human Behavior, 37(4), 303–314. tion of biological motion in newly hatched chicks as revealed by an Arterberry, M. E., & Bornstein, M. H. (2002). Infant perceptual and imprinting procedure. Animal Cognition, 3(1), 53–60. conceptual categorization: The roles of static and dynamic stimu- Ro, T., Friggel, A., & Lavie, N. (2007). Attentional biases for faces lus attributes. Cognition, 86(1), 1–24. and body parts. Visual Cognition, 15(3), 322–348. Brazelton, T. B., Koslowski, B., & Main, M. (1974). The origins of Rochat, P., Morgan, R., & Carpenter, M. (1997). Young infants’ sen- reciprocity: The early mother–infant interaction. In M. Lewis & L. sitivity to movement information specifying social causality. Cog- Rosenblum (Eds.), The effect of the infant on the caregiver. Wiley. nitive Development, 12(4), 537–561. Buiatti, M., Di Giorgio, E., Piazza, M., Polloni, C., Menna, G., Taddei, Rosa-Salva, O., Hernik, M., Broseghini, A., & Vallortigara, G. (2018). F., Baldo, E., & Vallortigara, G. (2019). Cortical route for facelike Visually-naı ¨ve chicks prefer agents that move as if constrained by pattern processing in human newborns. Proceedings of the a bilateral body-plan. Cognition, 173, 106–114. National Academy of Sciences, 116(10), 4625–4630. Salva, O. R., Mayer, U., & Vallortigara, G. (2015). Roots of a social Csibra, G., Gergely, G., B´ ıro, ´ S., Koos, O., & Brockbank, M. (1999). brain: Developmental models of emerging animacy-detection Goal attribution without agency cues: The perception of ‘pure mechanisms. Neuroscience & Biobehavioral Reviews, 50, reason’ in infancy. Cognition, 72(3), 237–267. 150–168. Di Giorgio, E., Loveland, J. L., Mayer, U., Rosa-Salva, O., Versace, E., Scholl, B. J., & Gao, T. (2013). Perceiving animacy and intentionality: & Vallortigara, G. (2017). Filial responses as predisposed and Visual processing or higher-level judgment? In M. D. Rutherford learned preferences: Early attachment in chicks and babies. & V. A. Kuhlmeier (Eds.), Social perception: Detection and inter- Behavioural Brain Research, 325, 90–104. pretation of animacy, agency, and intention (pp. 197–229). MIT Di Giorgio, E., Lunghi, M., Simion, F., & Vallortigara, G. (2017). Press. Visual cues of motion that trigger animacy perception at birth: The Simons, D. J., & Chabris, C. F. (1999). Gorillas in our midst: Sus- case of self-propulsion. Developmental science, 20(4), e12394. tained inattentional blindness for dynamic events. Perception, Di Giorgio, E., Lunghi, M., Vallortigara, G., & Simion, F. (2021). 28(9), 1059–1074. https://doi.org/10.1068/p281059 Newborns’ sensitivity to speed changes as a building block for Spelke, E. S., Phillips, A., & Woodward, A. L. (1995). Infants’ knowl- animacy perception. Scientific Reports, 11(1), 1–10. edge of object motion and human action. In Sperber, D., Premack, Fantz, R. L. (1963). Pattern vision in newborn infants. Science, D., & Premack, A. J. (Eds.), Symposia of the fyssen foundation. 140(3564), 296–297. Causal cognition: A multidisciplinary debate (pp. 44–78). Claren- Frankenhuis, W. E., House, B., Barrett, H. C., & Johnson, S. P. (2013). Infants’ perception of chasing. Cognition, 126(2), 224–233. don Press/Oxford University Press. Hofrichter et al. 9 Tra ¨uble, B., & Pauen, S. (2011). Infants’ reasoning about ambiguous Wagner, L., & Carey, S. (2005). 12-Month-old infants represent prob- motion events: The role of spatiotemporal and dispositional status able endings of motion events. Infancy, 7(1), 73–83. information. Cognitive Development, 26(1), 1–15. Wang, S. H., & Baillargeon, R. (2006). Infants’ physical knowledge Tra ¨uble, B., Pauen, S., & Poulin-Dubois, D. (2014). Speed and direc- affects their change detection. Developmental Science, 9(2), tion changes induce the perception of animacy in 7-month-old 173–181. infants. Frontiers in Psychology, 5, 1141. Woodward, A., Phillips, A., & Spelke, E. S. (1993). Infants’ expecta- Vallortigara, G. (2012). Aristotle and the chicken: Animacy and the tions about the motions of inanimate vs. animate objects. In Pro- origins of beliefs. In A. Fasolo (Ed.), The theory of evolution and ceedings of the Cognitive Science Society (pp. 1087–1091). its impact (pp. 189–199). Springer. Erlbaum.
Evolutionary Psychology – SAGE
Published: Jun 28, 2021
Keywords: animacy; change-detection; change-blindness; social attention
Access the full text.
Sign up today, get DeepDyve free for 14 days.