Poster group
Details of individual items:
poster
There are many theories of how infants categorize the world into animate agents and inanimate objects. A common proposal is that infants identify agents based only on characteristic patterns of motion (Mandler, 1992; Baron-Cohen, 1995; Premack, 1990; Gergely, et al,1995). The current research investigated the possibility that infants rely on other cues as well, in particular texture cues, which have been shown to be important for older children's identification of animates (Smith & Heise, 1992). We used the visual habituation paradigm to index infants' judgments of whether a stimulus was an agent. Previous work by Woodward (1998) demonstrated that infants selectively encode an actor's goal when they see a human arm reach for and grab an object but not when they see an inanimate object do so. We used this paradigm, but altered the texture of the human arm by having the actor wear a metallic gold glove. In a series of 6 studies, we found that infants of 7 and 12 months selectively attended to goals for normal, ungloved hands, replicating prior findings. Second, we found that at both ages, altering the arm's texture disrupted this pattern, suggesting that texture is an important cue to animacy for infants. Third, we found that pre-exposure to the gloved arm did not succeed in helping the infants to recognize it as part of an animate agent. Fourth, we discovered that infants' construal of the gloved hands could be altered by showing the babies the rest of the actor during the habituation procedure. That is, when they could see that the golved hand was part of a person, infants construed its actions as goal-directed. Taken together, these findings indicate that texture cues play an important role in infants' categorization of animates and inanimates, but that with enough information, babies can look past the texture cues when assessing animacy.References: Baron-Cohen, S. (1995). Mindblindness. Cambridge, MA: MIT Press. Gergely, G., Nadasdy, Z., Csibra, G., & Biro, S. (1995). Taking the intentional stance at 12 months of age. Cognition, 56, 165-193. Mandler, J. M. (1992). How to build a baby II: Conceptual primitives. Psychological Review, 99, 587-604. Premack, D. (1990). The infant's theory of self-propelled objects. Cognition, 36, 1-16. Smith, L. B. & Heise, D. (1992). Perceptual Similarity and Conceptual Structure. In Percepts, Concepts, and Categories. B. Burns (ed). Elsevier Science Publishers B. V., 233-271. Woodward, A. L. (1998). Infants selectively encode the goal object of an actor's reach. Cognition, 69, 1-34.
poster
The origins of infants' social skill are unknown, and current researchis inconclusive.Neonates prefer to look at facial configurations, and 3-6 month oldsrespond differently whenaspects of social interaction are violated (e.g., 'still face' effect),suggesting that young infantshave social expectations. Others find that tactile stimulation orcontingency carries the 'still-face'effect, suggesting a simpler explanation. Still others find that infantsrespond 'socially' whenmoving objects stop moving, become non-contingent, or become lessperceptually complex,suggesting that early social responses are not reserved for people.Questions thus remain aboutwhat recruits very young infants' social responses, and whether peopleuniquely do so. Existingwork has compared infants' gaze, smiles, vocalizations, and/or movements inresponse to peoplevs. objects. In the current study, in contrast, naive adults judgedwhether infants were interactingwith people vs. objects from videotaped interactions, using whatever cuesthe infants provided. 40 3-month-old infants participated in two experimental sessions. Ineach session, theyinteracted with 4 different 'partners' for 2.5 minutes each: mother; femaleexperimenter; face toy(black&white felt circle with eyes and nose); wiggleworm toy (brightlycolored commercial toy,6' long). Infants had been familiarized at home with one of the two toysbeginning at 1 month(half with worm; half with face). In the lab, all partners interacted at18' distance from the infant,without touch.. In one session persons and toys behaved like interactingpeople (Social Cuescondition: contingent; human voice; fluid movement), and in the othersession both behaved likemoving objects (Nonsocial Cues condition: noncontingent; squeak/rattle;rigid movement). Infants were videotaped in full-body view, without the partner in thecamera's view. Sixty trials were then randomly selected for a master tape, with thefollowing constraints:in 20 trials the infant was engaged with a person, in 20 with the face toy,and in 20 with the wormtoy; half the trials (30) were with a familiar partner (mother; familiartoy), and half unfamiliar(female experimenter; unfamiliar toy); half (30) featured social cues, andhalf nonsocial cues. Themiddle 45 seconds of each trial was copied, with order of trialsrandomized, and without sound(to ensure that observers were blind to condition). 35 university undergraduates made 3 forced-choice judgments about theinfant's partnerfor each of the 60 snippets: a) person vs. toy; b) familiar vs. unfamiliarpartner; c) social vs.nonsocial cues. When infants interacted with a person, the median rate ofcorrect person/objectjudgments was .55. When infants interacted with a toy, the median rate ofcorrect judgments was.53. Similar rates were obtained when infants interacted with familiar orunfamiliar partners (.50& .47 correct, respectively), and with partners exhibiting social vs.nonsocial cues (.53 & .43correct, respectively). These data suggest that naive adults cannotaccurately judge whether 3-month-old infants are engaged in interaction with another person or areresponding to nonsocialstimulation, on the basis of infants' facial expressions and movementquality alone.
poster
The foundation of infants' concept of animacy is thought by a number ofresearchers to be the motion properties of objects (e.g., Leslie, 1995;Mandler, 1992; Premack, 1990). Mandler (1992), for example, proposed thatinfants form rudimentary concepts - called image-schemas - that act as thebuilding blocks for infants' knowledge about the 'kinds of things' thatobjects are. These image-schemas are thought to embody the way that objectsbegin to move, the way objects move with respect to other objects, and thetrajectory that objects follow. As part of a more perceptually-orientedapproach, Rakison and Butterworth (1998) proposed that infants mayinitially acquire knowledge about the trajectory of animate and inanimatekinds by attending to the correlation between an objects' large, functionalparts and its movement path. Once this correlation is established, it isextended to the whole object that possesses the functional part. In two experiments, we used the habituation procedure to examine the notionthat infants are sensitive to specific correlations in the processing of amotion event. In Experiment 1, 14- and 18-month-old infants were habituatedto two events. In one event infants saw, for instance, a blue rectilineargeometric figure with yellow, horizontally moving parts that traveled alonga curvilinear path, and in the other event, infants saw a red curvilineargeometric figure with green, vertically moving parts that traveled along arectilinear path. Infants were presented with four events in the testphase, one identical to that seen during habituation, and three in which asingle property of one event - part, object, or motion path - was switchedwith the property of the other event. For example, in the part switch test,infants habituated to events described above would see a blue rectilinearobject with green, vertically moving parts that traveled along acurvilinear path. Results revealed that 14-month-olds dishabituated whenthe part was switched but did not dishabituate to the object or motion pathswitch. In contrast, 18-month-olds dishabituated in all three test trials.In Experiment 2, 14-month-olds were habituated to the same events as thoseused in the first experiment, with the exception that the parts no longermoved. The data revealed that infants did not dishabituate in any of thethree test trials. These data show that 14-month-olds attend to the correlation between partsand motion path and not to part-object, object-motion, orpart-object-motion correlations. Moreover, it is the movement, or dynamicaspect, of functional parts that makes this correlation particularlysalient to infants. As predicted by Rakison and Cohen (1999), infants mayfirst attend to part-motion correlations and then generalize thiscorrelation to include the part, the object, and the motion of an event. Inthis way, it is possible that infants learn that animates and inanimates -which differ in overall appearance and parts - follow distinct movementtrajectories. We suggest that a similar correlational mechanism might alsoaccount for infants' acquisition of knowledge about other aspects ofanimate and inanimate motion such as self-motion versus caused motion.
poster
The distinction between animate beings and inanimate objects is one thatsupports a range of inferences for adults. For example, knowing thatsomething is an animal allows us to infer that it is capable ofself-propulsion, growth, and reproduction (Gelman & Spelke, 1981). Theorigin of infants' understanding of the animate-inanimate distinction hasbeen the focus of a good deal of research and theory over the last decade.Most notably, Mandler (1992) has argued that infants develop conceptualcategories of animate and inanimate objects within the first 12 months oflife by abstracting motion characteristics of objects into image-schemas.There is little empirical evidence, however, to support the notion thatyoung infants develop a motion-related concept of animacy. Poulin-Dubois,Lepage and Ferland (1996) found that 9- and 12-month-old infants displayedmore negative affect to self-initiated motion by a robot than by a person,suggesting that they have different expectations for people and inanimateobjects. However, in a recent study of infants' understanding of motionproperties using the inductive generalization technique it was found thatinfants as old as 18 months are limited in their knowledge of the motionproperties of animates and inanimates (Poulin-Dubois, Rakison, Vyncke, &Baker, 1999).The present studies were designed to provide a further test of infants'knowledge about the motion capabilities of animates and inanimates.Specifically, we examined whether 12- and 16-month-olds understand thatanimals, but not vehicles, are capable of jumping over an obstacle. InExperiment 1, animated films were presented to twenty-four 12-month-oldson a computer screen using the habituation procedure. In the habituationphase, infants were presented with one event in which a dog jumped over awall and one event in which a car hit the wall and bounced. In the testphase, infants were presented with two films: In the Congruent film, anovel animal or vehicle engaged in the motion demonstrated by a member ofthe same category in the habituation phase (e.g., a cat jumped over thewall). In the Incongruent film, a novel animal or vehicle violated theexpected motion path (e.g., a bus jumped over the wall). Infants lookedequally long at the two test events, suggesting that they have not yetassociated animals and vehicles with these particular motion events. Todetermine the age at which this knowledge develops, a second study wasconducted with 16-month-olds using the same procedure. Preliminaryresults (N15) reveal that 16-month-olds also do not understand the motionevents tested.The results of these studies suggest that infants' knowledge of animateand inanimate motion does not emerge as early as suggested previously.This may reflect that infants do not understand the motion differencesbetween animate and inanimate objects until after 16 months of age, orthat an understanding of certain motions (e.g., self-propulsion) developsbefore others. We are planning to report additional studies, one with20-month-old infants using the films discussed here, and one in whichadditional motion cues for animate and inanimate objects are provided(e.g., self-propulsion and contingency).
poster
Infants possess sophisticated categorization abilities that enable them tomake such discriminations as animals from vehicles and humans fromfour-legged animals (Mandler & McDonough, 1996; Pauen, in press; Ross,1980). Various sources of information have been proposed to account forinfants' categorical distinctions. For example, motion cues have beenposited to underlie the ability to discriminate animate from inanimateobjects in 9-month-olds (Poulin-Dubois, Lepage, & Ferland, 1996). Inaddition, research has shown that 7-month-olds have different expectationsabout the movement behavior of animate and inanimate objects (Woodward,1997). Findings such as these raise the question of how infants acquiretheir knowledge of categories. How readily are infants able to learnfeatural information relevant to the categorization of individual objects? To address this question, 7-month-old infants were familiarized toa pair of wind-up toy animals as they moved across a stage. The animalswere selected to be perceptually distinct from one another in shape, color,and texture. First, infants were presented with six familiarization trialsin which the toy animals were lowered onto a stage by a human hand and thenmoved across the stage, for a total of three trials per animal. Theprimary difference between the two animals was its method of movement: oneof the animals exhibited self-propelled (SP) motion, whereas the other washand-moved (HM) by the experimenter. The SP animal was lowered, releasedfrom the hand, moved across the stage for 8 s, and then held motionless onthe stage by hand for 4 s. The HM animal was lowered, moved by hand acrossthe stage for 8 seconds, and then released from the hand remainingmotionless on the stage for 4 s. Infants were then presented with two testtrials in which both animals were simultaneously placed onto the stage andremained motionless for 30 s, with their lateral positions reversingbetween trials. The initial side of placement of the two animals and theparticular animals assigned as SP vs HM were counterbalanced. Using apreferential looking procedure, the amount of time each infant spentlooking at each of the two animals for the duration of the two 30-s testtrials was recorded. The entire procedure was then repeated with two newwind-up toy animals.If infants map features of objects after limited exposure, then they shouldlook significantly longer at the animal which moved on its own (SP) duringthe familiarization trials compared to the other animal. This predictionis motivated by the expectation that the self-propelled animal will beginto move once it is released, whereas the hand-moved animal will remainmotionless. The results confirmed this prediction with 14 out of 17infants looking longer at the SP animal across the four test trials (p <0.02). An ANOVA revealed a main effect of type of movement (SP vs. HM)[F(14) 12.97, p < .005], with infants looking significantly longer at theSP animal compared to the HM animal (15.98 s compared to 11.67 s,respectively). In subsequent studies we are investigating infants'long-term retention of mappings between specific objects and theirfeatures, and how they generalize this knowledge to new objects. Theresults will be discussed both in terms of infants' developinganimate-inanimate distinction and infants' developing capacity to learn andremember arbitrary mappings between specific objects and their features.
poster
The present studies investigated the sources of information infants use to individuate a visual display into objects, particularly with respect to individuating people. Previous work suggests that by 9 months infants can individuate objects on the basis of spatiotemporal information (e.g., common movement; Spelke & Kellman, 1983) and general featural information (i.e., the coincidence of color and shape boundaries; Needham, 1998) but not kind-specific featural information (i.e., the features that distinguish the boundaries of a class of objects; Xu, Carey & Welsh, 1999). Our studies examined whether infants' inability to use kind-specific information to individuate objects extended to the highly familiar kind person; specifically, whether infants infer the boundaries of the human body in lieu of spatiotemporal or general featural information.Forty-one 9-month-olds were tested in two conditions, Spatiotemporal and Kind-specific, using a standard habituation paradigm. Infants viewed a human arm sitting on top of a soft, flesh-toned, cylindrically-shaped object on a puppet stage. During habituation in the Spatiotemporal condition, the arm moved independently of the other object. The display was static throughout habituation in the Kind-specific condition. In both conditions, test trials consisted of alternating events in which the arm moved upward, and the second object either remained behind on the stage (expected event) or moved upward with the arm as if they were one object (unexpected event). Infants were credited with dividing the display into two objects if they looked longer at the unexpected event than the expected event, relative to a baseline preference of other infants in this context to look at the two separate objects (N ). Repeated measures ANOVAs indicated that infants in the Spatiotemporal condition individuated the two objects, F(1,35)4.46, p< .05, but infants in the static Kind-specific condition did not, F(1,38)1.85, ns.A second study investigated the possibility that infants were simply inattentive in the static Kind-specific condition. Eighteen additional infants were habituated to a static condition in which the arm rested on a bright blue, angular box (General Featural condition). In all other respects, the test events were identical to the previous study. Infants in this condition looked longer when the two objects moved as one, in comparison to baseline, F(1,32)4.08, p.052. Further comparisons confirmed that the Kind-specific condition differed from both the Spatiotemporal and General Featural conditions, p's <.05, which did not differ from each other. Nine-month-olds do use static features to individuate a display; however, they use only those features that indicate object boundaries regardless of object kind.These findings suggest that infants individuate the human body as they do other physical objects, using spatiotemporal and general featural information, but do not use the featural information inherent to the kind person to do this same task. These findings are somewhat surprising in light of evidence that infants have different expectations about the human body than other objects (Leslie, 1984; Woodward, 1998). However, they support the hypothesis that, in the infants' first year, only the concept solid physical object provides the criteria for individuation (Xu & Carey, 1996). ReferencesKellman, P.J. & Spelke, E.S., (1983). Perception of partly occluded objects in infancy. Cognitive Psychology, 15(4), 483-524.Leslie, A., (1984). Infant perception of a manual pick-up event. British Journal of Developmental Psychology, 2, 19-32.Needham, A. (1998). Infants' use of featural information in the segregation of stationary objects. Infant Behavior and Development, 21:47-76.Woodward, A.L., (1998). Infants's selectively encode the goal object of an actor's reach. Cognition, 69, 1-34.Xu, F. & Carey, S. (1996). Infant's metaphysics: the case of numerical identity. Cognitive Psychology, 30:111-153.Xu, F., Carey, S. & Welch, J. (1999). Infant's ability to use object kind information for object individuation. Cognition, 70:137-66.