Poster group
Details of individual items:
poster
no abstract
poster
This research explored infants' use of prior experiences with objects toparse an otherwise ambiguous display. In previous research, it wasshown that, when presented with a test display consisting of a tall,blue box and a curved yellow cylinder, 4.5-month-old infants were unableto determine the composition of the display. However, certain priorexperiences seemed to allow infants to segregate the test display intotwo separate units. In one set of studies it was shown that seeingeither the test box or the test cylinder alone prior to testingfacilitated infants' subsequent parsing of the test display. Similarly,prior exposure to a collection of three boxes, each of which wasnoticeably different from the test box, also facilitated infants'perception of the display. However, seeing three identical copies ofone of these boxes did not facilitate infants' parsing of the testdisplay. One explanation for these findings is that seeing the threedifferent boxes allowed infants to form a category of boxes generalenough to include the test box, and that they then used thisrepresentation to segregate the test display into two separate objects. In the current research, follow-up studies were performed to explore theeffects of seeing two of the three different boxes prior to testing(Experiment 1), and seeing two boxes that each had some of the testbox's features (Experiment 2). In both experiments, the infants saw twoboxes prior to testing that resembled the test box in certain ways. Theonly difference between the studies was in the specific ways in whichthe familiarization boxes resembled the test box. The infants in Experiment 1 saw different pairings of two of the threedifferent boxes shown to facilitate infants' parsing of the display inprevious research (either boxes 1 and 2, boxes 2 and 3, or boxes 1 and3). The results of each of these conditions indicated that infants'segregation of the test display was not facilitated by these priorexperiences. Together with previous findings, these results suggestthat in this context, exposure to three different boxes may be theminimum conditions necessary for infants to form a general categoricalrepresentation that is broad enough to include the test box. The infants in Experiment 2 received a prior exposure to two boxes thateach had some features that were the same as the text box and some thatwere different. Between the two boxes, all of the test box's featureswere represented. Although prior exposure to either one of these boxesalone did not facilitate infants' parsing of the test display,simultaneous exposure to both boxes did facilitate infants' perceptionof the test display. These results suggest that the infants may havecombined the features of the two boxes and experienced an illusion ofhaving seen a single object with the features of the test box. Likeadults, infants may experience 'illusory conjunctions' in which thefeatures of two objects are incorrectly combined.The results of both of these experiments provide information about therepresentations created in infants' short-term visual memory. Thesephenomena are important to study if we want to gain a betterunderstanding of the process of object recognition during infancy.
poster
In children older than 2 years, the processing of objects' properties is mainly studied through categorisation tasks (grouping of objects depending on their similarities), as categorisation may be considered as a fundamental aspect of knowledge. According to a number of works, the dissociation of objects' properties, which is a pre-requisite to categorisation, develops slowly. For instance, Piaget and Inhelder (1959) specified a number of steps for the construction of classes, while Wallon (1945) described how a syncretic mode of thought shifts to a categorical one. Last, Mounoud (1986) defines a change from a perceptual representation to a categorical one. In order to clarify the processes underlying such shifts, authors put forward a number of factors : perception ' knows ' similarity relations, and partitive membership ; action leads to operations which involve inclusion relations; comparison and analysis processes which induce the abstraction of objects' properties. In a paper dated 1950, Wallon and Ascoli note that the property of shape is more difficult to isolate than the property of colour. They suggest that shape has a specific power for categorisation. More recently, Weil-Barais (1999), in a research about teaching scientific concepts to preschoolers, observes that although children use words to name shape (rectangle, square, circle, triangle, etc.), they do not master the generic concept of shape.Following Mounoud's theory, we aim to study the elaboration of the concept of shape through a motor fitting task. Mounoud (1986) stresses the interaction between motor and cognitive skills : 'because we experiment through our actions (empirical regulation), we elaborate new knowledge ; conversely because we know things our actions are planned in advance (abstract representation)'. We aim at demonstrating how 2- to 5-year-olds progressively abstract the notion of ' shape ' through the development of their motor skills : a fitting task may be more or less rapidly performed, and more or less successfully performed.We hypothesize that for 3-to-4-year-old children, an action must be performed for the notion of shape to be elaborated (empirical regulation), while from 5 years on, this property is already acquired, and makes it possible to plan the performance and guide action. In order to outline this development, we observed 80 children aged 2-, 3-, 4- and 5-years in two types of tasks : the first one requests actions (fitting task) while in the second one, children have to point to the right fitting place without manipulating the objects.Children were videotaped in three conditions : in the matching task, objects have the same shape and colour as the fitting peg; in the non-matching task, objects have the same shape but a different colour; in the neutral condition objects and fitting places are not coloured. Subjects are asked either to point to the peg which has the same shape as the piece, or to fit the piece on the corresponding peg.In the fitting situation, 4-year-olds are more successful than 2- and 3-year-olds, particularly when colour competes with shape, but need more time to elaborate this response. Only in 5-year-olds are all three conditions totally and rapidly mastered. In the pointing situation, 5-year-olds have still difficulties in the non-matching task. This suggests that the abstraction of the notion of shape is still being elaborated.
poster
Recent research suggests that young children categorize objects by usingboth "obvious" perceptual cues such as shape (Smith et al.,1996), and "non-obvious" cues such as function, name or causalproperties (Gelman & Markman, 1986, 1987; Gelman & Coley, 1990;Gopnik & Sobel, in revision; Kemler Nelson, 1995). These studiesfurther suggest that obvious cues are used earlier than non-obvious ones.The present study, following up on Gopnik and Sobel (in revision),explores the relative weighting of perceptual and causal cues tocategorization for 3.5- and 4.5-year-olds.
An object manipulation method was used. For each trial, children werepresented with four objects, which were placed on a machine one at a time(while giving linguistic descriptions). Two objects made the machinework, that is light up and play music, two did not. Then, theexperimenter labeled one of the objects that had made the machine work a"tib," and asked the child to give him the other"tib" (label extension question). When children chose the
"incorrect" object as the "tib" -- that is chose anobject that did not make the machine work -- they were given a memorycheck. In the memory check, children were simply asked which object hadmade the machine work. Each child was tested on six different conditions,as three linguistic descriptions and two types of trials were used. Thelinguistic descriptions were either causally biased ("Look, it makesthe machine work!"), perceptually biased ("Look, it'sred!"), or neutral ("Look, the machine works!"). In theno-conflict trials all the objects were different, so that the only cueto categorization was the causal properties of the objects. In contrast,in the conflict trials two pairs of identical objects were used. Oneobject of each pair made the machine work, resulting in conflictingperceptual and causal categorizations. This allowed a direct comparisonof children's relative weighting of both cues.
The results first confirm that children can use causal cues tocategorization at both ages. They also establish an age-related shift inthe relative weighting of perceptual and causal cues. When the cuesconflicted, the younger children mainly relied on perceptual cues and theolder ones on causal cues. Importantly however, the younger children whoused the perceptual cue to determine which other object was a"tib" also misremembered the causal properties of that object:they said incorrectly that the perceptually identical object had made themachine work. This suggests that 3.5-year-olds use perceptual cues notonly as a guide to names but also as a guide to causal properties (inspite of available counter evidence), while 4.5-year-olds have discoveredthat both cues can be dissociated. Younger children apparently expectnames, causal properties and perceptual properties to be stronglycorrelated while older children recognize that these properties may be inconflict. This interpretation is further supported by the finding thatcausal language only influenced the categorization behavior of the olderchildren.
poster
no abstract
poster
Interrelationships between cognitive and linguistic developmenthave long been of interest to researchers in both fields. Indeed, measuresof vocabulary size and verbal subscales of intelligence tests are oftenused as outcome variables for earlier cognitive predictor variables. Butas noted by Shore, Dixon, and Bauer (1995), it's important that measures ofcognitive development not be completely confounded with measures oflinguistic development. In the present paper we consider the extent thatearly measures of cognitive functioning are predictive of laternonlinguistic cognitive measures. Thus, 7-month measures of habituationare used to predict both 20-month conceptually-based sequential touchingand 20-month elicited symbolic play. Approximately 30 babies and their mothers visited the laboratory atchild ages 7 and 20 months. Not all babies completed all tasks, so aslightly varying number of babies were involved in the various predictionsdescribed below. The children came from a variety of socioeconomicbackgrounds, but were primarily European-American. Habituation measureswere obtained using a modified infant-controlled fixed interval procedurein which the infant controlled the number of trials but not trial onset oroffset. The stimuli included six color photographs of common objects foundin children's books. Test trials consisted of novel stimuli from the samebooks. Habituation measures included trials to criterion, number of peakfixations, and trial of first peak fixation. Sequential touching measureswere extracted from children's performance on a modified version of thesequential touching task (Mandler, Fivush, & Reznick, 1987). On this task,children were presented with trays containing six stimulus objects fromeach of two conceptual categories. The categories represented were eithervehicles, foods, or household objects. When possible, children werepresented with all three combinations of conceptual category contrasts.Children's best performance in distinguishing between conceptual categorieson any one tray, and their mean performance across all trays was assessed.Symbolic play performance was measured using an elicited imitationtechnique (Bauer & Shore, 1987). After first manipulating the props,children saw a 'making-breakfast' and a 'going-to-bed' scene modeled twice.Each scene comprised four constituent acts. Children were then permitted toengage the props a second time. Improvement over spontaneous performancein the total number of acts, longest chain of acts, pairs of acts in properorder, and variety of acts was assessed. The 7-month habituation measures were found to be predictive ofboth conceptually based sequential touching and improvement overspontaneous pretend in the going-to-bed scene (Table 1). Predictions to'making breakfast' performance were not significant. The fact that children's visual information processing prior to thefirst birthday was predictive of other putatively cognitive measures morethan one year later suggests that there may very well be a cluster ofnonlinguistic indicators of cognitive functioning. This interpretation isbuttressed by evidence from our lab that 7-month habituation measures arenot correlated with 20-month vocabulary productivity. The finding thatimprovement in bed play, which involved pretend on an animate other, andnot breakfast play, involving pretend on the self, suggests a differentialinvolvement of pretend play in nonlinguistic cognition as a function ofplay content.Table 1Correlations Between 7-Month Habituation Measures and 20-Month SequentialTouching and Elicited Symbolic Play Measures HabituationMeasures20-Month Measures Trials to Criterion # Peak Fixations Trial 1st PeakMean Categorizing .55* .60*Best Categorizing .47* .73*Improvement from Spontaneous to Elicited Total Pretend Acts (Bed) .40* Longest Chain Acts (Bed) .57* .37 .48* Pairs of Acts (Bed) .42* Variety of Acts (Bed) .49* .48* .64*Note: * p < .05, else p < .10.References Bauer, P.J., & Shore, C.M. (1987). Making a memorable event:Effects of familiarity and organization on young children's recall ofaction sequences. Cognitive Development, 2, 327-338. Mandler, J.M., Fivush, R., & Reznick, J.S. (1987). The developmentof contextual categories. Cognitive Development, 2, 339-354. Shore, C., Dixon, W.E., Jr., & Bauer, P.J. (1995). Measures oflinguistic and nonlinguistic knowledge of objects in the second year.First Language, 44, 189-202.
poster
Previous research has indicated that object labels influence2-1/2-year-oldsB9 inferences about nonobvious object properties (e.g.,Gelman & Coley, 1990), however little is known about whether infantswill use labels as an inductive base. Although infants can use sharedshape similarity to make inferences about objectsB9 underlying sharedproperties (e.g., Baldwin et al., 1993), the degree of shape similarityrequired for infants to make these inferences is still relativelyunknown. The goals of the current study were to investigate: (1)whether infants will use object labels to guide their inferences aboutnonobvious object properties, (2) how infants use shape similarity as ameans for making inductions, and (3) whether age and productivevocabulary size impact infantsB9 ability to make such inferences.Sixteen-to-twenty-one-month-olds were presented with novel objects intwo between-subjects conditions. In the Salient Label condition,infants were provided with a novel label for each object when it waspresented (e.g., 'Look at this flum!'). In the No Label condition,infants were not taught labels for the same objects (e.g., 'Look at thisone!'). All participants saw target prototype objects and correspondinghigh, medium, and low shape similarity test objects in two conditions ofinterest. In the Surprised condition, a target object was presented(e.g., a disguised ball) and its nonobvious property (e.g., squeaks whensqueezed) was demonstrated by the experimenter. After exploring thisobject for 10 seconds, infants were presented with a corresponding testobject that was disabled so that it could not make the target sound(e.g., would not squeak) and explored it for 20 seconds. In theInterest Control condition, both the target and test object presented(e.g., disguised rattles) were disabled and did not make a target sound(e.g., would not rattle). A comparison of the frequency of targetactions performed on test objects indicated that infants performedsignificantly more target actions in the Surprised Condition than in theInterest Control condition, indicating that the objectsB9 properties wereindeed nonobvious and that infants formed expectations and madeinferences about shared object properties on the basis of experiencewith a prototype exemplar. We also found that infants performedsignificantly more target actions on high shape similarity objects thanon medium and low shape similarity objects. Finally, infants performedsignificantly more target actions in the Salient Label condition than inthe No Label condition at all levels of shape, regardless of age orproductive vocabulary size. Of particular interest was the finding thatinfants performed significantly more target actions on the lowsimilarity object in the Salient Label condition in comparison to the NoLabel condition. This finding indicates that infants can rely on labelsto make inductive inferences, in the absence of shared shape similarity.The results of this study provide important insights into earlyinductive abilities, demonstrating that by 16 months of age, shape andlabel information will promote inductive inferences about nonobviousproperties of novel objects.
poster
Contextual and taxonomic categorisations are explored in the second year of life, with an object manipulation task used to assess the development of categorisation in 3 age groups at 11, 15 and 18 months, and then for a group seen every 4 weeks. An interaction with a picture book is also videotaped, at the end of every session. The cross-sectional study, as well as the follow-up one, shows that contextual categories appear neither earlier than taxonomic ones, nor easier. Taxonomic categories do not appear as hierarchically learned. No age difference is observed. Important individual differences are observed at any age, and specially at 15 months of age. Individual consistency will then be investigated, according to the different ontological categories and to parental use of distancing acts during the picture book session.
poster
Several researchers have argued in recent years that a weak form of theWhorfian hypothesis may be sustained in certain areas of cognitivedevelopment (de Villiers, 1997; Spelke & Condry, 1999). The presentstudy investigates the impact of labeling on infants' representationsof kind concepts such as
poster
Pictures are both objects in their own right and symbols of whateverthey depict. Their resemblance to what they represent has led to theassumption that the symbolic nature of pictures is easily understood, evenby young infants. A substantial body of research has demonstrated thatinfants can recognize pictured objects and people, but recognition is notequivalent to understanding the nature of these representations. Indeed,our research suggests that infants do not fully understand the symbolicnature of pictures. In previous research we systematically investigated 9-month-oldsinfants' response to color photographs of objects. In several studies,most subjects have manually investigated photographs presented to them inbooks. The infants have rubbed and hit the pictured objects, and in manycases grasped at them, as if trying to pluck them off the page. Theinfants' behaviors are surprising in light of the well-established abilityof even younger infants to discriminate between 2-dimensional and3-dimensional stimuli. The interpretation offered posits that infants areexploring pictured objects because they are unsure about the nature ofpictures: In some ways pictures look like real objects, but in other waysthey do not. Infants manually explore pictured objects in order toinvestigate this apparent paradox. In the research presented here we examine whether distinctivecharacteristics of the objects depicted in pictures affect infants' manualinvestigation of them. A group of 9-month-old infants was presented with aseries of 12 objects and 12 pictures of those objects. All of the objectshad handles. Half of the objects had salient texture opposite the handle(e.g., feathers, brush bristles, fuzzy pieces of sponge), and half had asmooth surface on the opposite end (e.g., sphere, spoon, ladle). (SeeFigure 1). Pictures were presented individually using a speciallyconstructed display table. Indeed, infants' manual investigation of the pictures differedaccording to the nature of the depicted stimuli. Infants tended to rub atdepictions of texture and did so significantly more than at the depictedhandles. Conversely, infants grasped more at the handles than the ends,though this difference was not significant. (See Figure 2.) Infants'behavior to the actual objects was very similar: Infants explored the end,but were more likely to grasp the handle. Thus, infants rubbed thedepicted texture as though trying to experience the salient fuzzy, bristly,or feathery texture; they grasped as if to pick up the depicted objects bythe handle. (See Figure 3.) This study also confirms that infants' manualinvestigation is reliable and replicable, evident when pictures arepresented individually as well as in books. This research supports the position that 9-month-old infants manuallyinvestigate pictures because they do not understand how 2-dimensional and3-dimensional stimuli differ. As a result, infants respond to picturedobjects in terms of the properties of the actual referents. With continuedexperience with both objects and pictures, infants come to fully grasp therepresentational nature of pictures.