Symposium
Chair: David H. Rakison
Discussant: Laraine McDonough
The development of a concept of animacy is thought to be one of thecornerstones of cognitive development. Not only does it entail the abilityto categorize objects into distinct kinds - animals, vehicles, plants, andtools, for instance - but it also involves the attribution of certainproperties to members of those different kinds; for example, animals arealive, vehicles have motor engines. However, though a large database existsconcerning preschoolers' and older children's understanding of theproperties of animates and inanimates (e.g., Carey, 1985; Keil, 1991;Gelman & Markman, 1986, 1987), relatively little is known about thedevelopment of a notion of animacy in infancy (see Poulin-Dubois, 1999). The research that does exist on infants' notion of animacy has focused ontheir perception and understanding of two aspects of animates andinanimates, namely, psychological properties and physical propertiesrelated to motion. Because of the complex, multifaceted nature of animatesand inanimates, these properties can be analyzed further into a number ofcharacteristics. Motion properties can be described, for example, in termsof self-starting versus caused motion, contingent versus noncontingentmovement, and linear versus nonlinear motion paths (Mandler, 1992; Premack,1990), whereas psychological properties are often discussed in terms ofintentionality and goal-directedness (Meltzoff, 1995; Zeedyk, 1996). Atissue is how and when infants discriminate among each of these properties -for example, distinguishing the agent from the recipient of an action - andassociate particular object kinds with different aspects of each property;for example, animals, but not artifacts, are self-starting. Of particularinterest is whether infants develop a coherent concept of animacy thatincorporates together the different properties of animate and inanimateentities; for example, entities that are self-starting also tend to begoal-directed, move nonlinearly, and so on. The purpose of this symposium is to allow four research teams that focus ondifferent aspects of infants' notion of animacy to present their mostrecent findings. Because each research team focuses not only on differentaspects of the animate-inanimate distinction but also on different agegroups within infancy, the symposium aims to provide an overview of thedevelopment of the notion of animacy. Two presentations report studies onthe role of social cues in the development of the animate-inanimatedistinction. One presentation examines how the detection of independentsocial causality is rooted in early dyadic experience in which infantslearn to appreciate social partners as communicative agents, and a secondreports on whether infants expect people to interact differently withsocial and nonsocial entities. Two additional presentations focus oninfants' perception and interpretation of motion cues. One focuses onwhether infants understand the relationship between self-propulsion andgoal-directedness, and the other reports on infants' ability to makeinferences about the motion events in which animates and inanimates engage.The presentations will be discussed in terms of the coherence of infants'concept of animacy and the role of perceptual cues and conceptual knowledgein the development of this concept.
Details of individual items:
paper
In the broad sense, animated entities are things that 'breath life',whether these entities are really alive, or seemingly so as in the case ofpuppets, cartoon figures, or mechanical artifacts that are surreptitiouslyset in motion. But what makes some things to be perceived and construed asbreathing life, and other not? In this paper, I argue that early on, theanimate-inanimate distinction rests on the basic discrimination betweenentities that are perceived as behaving either toward a goal, or accordingto intrinsic plans, and all the other entities that behave either randomlyor passively under external circumstances (e.g., rolling of a kicked ball).The general ideas developed here are that the origin of theanimate-inanimate distinction in infancy is inseparable from 1) the originof an intentional stance starting to be manifested by infants at around 10months of age, 2) the development of a sense of self agency developing frombirth, but emerging unambiguously at around 2 months of age, and 3) theprogressive mapping of the primary sense of self-agency onto non-selfentities, including others, starting in the second month of life. Empirical evidence are presented suggesting that from birth and in thecourse of the first 8 weeks of life, infants develop a sense of themselvesas animated entities. They engage in the systematic exploration of theperceptual consequences of their own action. In the context of activeself-exploration, young infants develop a perception of themselves asagents situated in the environment, and as such, they are animated by plansand goals, and they do not act randomly. This early sense of self-agency isviewed as foundational of the animate-inanimate distinction that infantseventually develop in relation to other, non-self entities.The question of how infants, from developing a sense of self agency, mightperceive agency in other people as well as in animated objects isdiscussed. Early face-to-face dyadic exchanges are considered as a uniquelyfertile, highly scaffolded terrain for infants to develop an ability to mapand match their own sense of animacy with the animacy of others. Earlydyadic exchanges are considered as a determinant of the mapping between thesense of self-animacy developing from birth in the context ofself-exploration and the developing sense of animacy in non self entitiesevident beyond the second month of life. In support of this claim,empirical evidence are provided pointing to the parallel development ofsocial competence and of animate-inanimate discrimination of abstractfigures moving on a computer screen.
paper
Before infants can understand that other persons are intentional beings,that have goals that can be achieved with different means, they mustunderstand that people are animate beings, that not only move by themselvesbut also behave in certain ways. However, in order for infants to be ableto understand that others are animate agents they must understand that theythemselves are animate agents that make things happen (Tomasello, in press,Wellman, 1993). From birth, infants identify with other humans (Trevarthen, 1979; Bruner, 1983), and by4 months, infants call for hidden people and reach for the occluder behindwhich objects have disappeared (Legerstee, 1994). This suggests thatinfants understand that people can be acted on from a distance, but thatinanimate objects need to be contacted. In the current study we askedwhether 5- to 6-month-old infants expect other people to behave in waysthey do. In Experiment 1, infants were habituated to an actor who eithertalked to, or reached for something hidden behind an occluder. In the testevents the occluder was removed, and the infants saw the actor reach for,or talk to, either a person or an inanimate object. Infants looked longerwhen the actor talked to or reached for the object than the person. InExperiment 2, we tested whether infants' differential responsivenessreflected test rather than habituation experiences. Infants were habituatedas before, but were randomly assigned to test conditions in which theperson or the inanimate object was seen in a stationary position. Infantsthat had been habituated to an actor talking, looked longer at an inanimateobject than a person during the test events. Infants that had been habituated to an actor reaching, looked longer to a person than to aninanimate object during the test events. These interactions show thatinfants are surprised when they see people performing actions in which theywould not usually engage. These findings suggest that 5- to 6-month-oldinfants understand that other people communicate with social objects butact on inanimate objects.
paper
The proper domain of naive psychological reasoning is human action andhuman mental states but such reasoning is frequently applied to non-humanphenomena, such as natural forces or behaviour of computers as well. Theusual explanation for such a tendency holds that it reflects anoverextension of an interpretational strategy, which has originallydeveloped for human intentional actions. However, several studies havedemonstrated that infants are willing to attribute intentionality tonon-human objects as well. Therefore, as an alternative to the strictlinkage of psychological reasoning to human beings, several authors haveproposed that the initial domain of the applicationof psychological principles corresponds to the wider ontological categoryof animates or agents where the category membership is defined by theobject s ability to initiate movement by itself. In the present studies wetested the validity of the hypothesis that goal-attribution is restrictedto, and triggered by, the perception of self-initiated movement. In thefirst experiment, 9- and12-month-old infants participated in a habituation procedure similar to theGergely et al. (1995) study. In the habituation phase of the experimentalcondition the infants observed a computer-animated figure approachinganother one by flying over an obstacle of varying height. Since the heightof the movement trajectory was adjusted to the height of the obstacle, theobject always appeared to take the shortest, most efficient pathway. At thesame time, since it entered the screen already flying and moved in an inertmanner, it displayed no cues of agency such as self-propulsion. Infants inthe control condition saw the same habituation events with the differencethat the objectgot to its target position by flying over a bar hanging in the air anddid not take the short, straight route under the bar to the goal object. Inthe test phase (which was identical in both conditions), the obstacle orthe hanging bar was removed and the object took either the shortest,straight horizontal path to the goal-object, or flew there following theparabolic trajectory. In both age groups infants in the experimentalcondition looked longer at the flying object than at the horizontallymoving object (p<.01), while in the control condition no such differencewas found. These results demonstrate that the infants in the experimentalcondition could recognise the rationality of the object s goal-approachingbehaviour in the absence of evidence of self-initiated movement and couldpredict its future action accordingly. A second control study (in which thetarget object was stationary) ruled out a possible alternativeinterpretation of this finding according to which agentive properties mayhave been attributed to the flying object through generalization from thesimilar looking target object which in the first study was self-propelled.These results suggest that the domain of naive psychology is initiallydefined only by the applicability of its core principles and its ontologyis not restricted to featurally identified object kinds such as persons,animates, or agents.
paper
According to a number of researchers, one of the best contenders for thefoundation of infants' notion of animacy is the motion characteristics ofobjects (e.g., Leslie, 1995; Mandler, 1992; Poulin-Dubois & Shultz, 1988;Premack, 1990). This is perhaps not surprising given that infants aresensitive to motion virtually from birth (e.g., Slater, 1989) and thatchildren and adult believe that motion is one of the most importantcriteria for judging whether a novel entity is animate (e.g., Richards &Siegler, 1986). However, although a large number of studies haveestablished that infants in the first year of life discriminate betweenanimate and inanimate related motion characteristics - for example, agentfrom patient roles in a causal scene - few experiments have examinedwhether infants understand that animate and inanimate objects take part indifferent motion events.In two experiments, we used the inductive generalization procedure to test14- and 18-month-old infants' knowledge of motion events common to animateand inanimate objects. In Experiment 1, we showed infants two motionsappropriate to animals (e.g., climbing stairs) and two motions appropriateto vehicles (e.g., jumping from one ramp to another). A dog and a car wereused for modeling the motion events. Generalization was tested usinganother animal (e.g., rabbit) and another vehicle (e.g., bus). To establisha baseline measure, infants were given the test exemplars and the prop(e.g., stairs) before the motions were modeled. Although infants' overallnumber of responses were low, results revealed a significant increase inthe number of properties produced with the target category from baseline togeneralization at both ages.In Experiment 2, we tested whether infants' behavior in the firstexperiment was driven by knowledge of the motion associated with animateand inanimates or by a process of imitating the experimenter's actions withthe most perceptually similar exemplar. The same events and stimuli wereused, but each event was modeled with the distractor category exemplar(e.g., using a car to go upstairs). Results revealed that 14-month-oldinfants were no more likely to model events with a member of the distractorcategory than a member of the appropriate (target) category, though thenumber of responses with the distractor category members were higher thanthose in Experiment 1. The low overall response rate found in Experiment 1 suggests that infantsin the second year of life have only begun to associate certain objectkinds with certain motion events. The imitation of motion events with bothappropriate and inappropriate exemplars in Experiment 2 suggests thatinfants' behavior in these inductive generalization tasks may have beenguided as much by a process of perceptual matching and imitation than byconceptual knowledge. Drawing from recent research in our lab with thehabituation paradigm, we propose that infants develop knowledge about themotion capabilities of animates and inanimates through observation ofcorrelations between objects' appearances and their movement patterns.