Poster group
Details of individual items:
poster
How during development do we come to know which edges and contoursgo together to form which objects? Gestalt psychologists have argued thatour nervous systems are constrained to follow certain grouping principles(e.g., closure, common movement, good continuation, proximity, andsimilarity) that specify how small pieces of a visual scene should beorganized to form larger perceptual units or perceptual wholes. Quinn, Burke, and Rush (1993, IBAD, 16, 19-42) provided empiricalevidence relevant to the Gestalt account of the development of objectperception by investigating whether 3-month-old infants could organizevisual patterns in accord with the grouping principle of lightnesssimilarity. The infants were familiarized with arrays of elements thatadults perceived as rows or columns based on alternation of the lightnessvs. darkness of the elements. The infants were then given a noveltypreference test that paired horizontal vs. vertical stripes. The resultswere that infants familiarized with columns preferred horizontal stripes,whereas infants familiarized with rows preferred vertical stripes. Thesefindings indicated that young infants were able to use lightness similarityto represent the row- vs. column-like organization of the arrays ofelements. In the current experiment, we asked whether infants could organizevisual patterns in accord with form similarity. Two age groups of infants,3- to 4-month-olds (N 32) and 6- to 7-month-olds (N 32), werefamiliarized with arrays of elements consisting of alternating rows orcolumns of Xs and Os during 6, 15-s trials, and then given a noveltypreference test pairing horizontal vs. vertical stripes during 2, 10-strials. Mean looking times during familiarization for trials 1-3 and trials4-6 were 10.65 s (SD 2.88) and 9.38 s (SD 3.53) for the 3- to4-month-olds, and 7.51 s (SD 2.68) and 6.11 s (SD 2.48) for the 6- to7-month-olds. A three-way ANOVA, age (3 to 4 months vs. 6 to 7 months) xfamiliar stimulus (columns vs. rows) x trials (1-3 vs. 4-6), revealed bothan effect of trials, F(1, 56) 26.24, p < .001, and an effect of age, F(1,56) 22.95, p < .001. These effects indicated that both groups of infantshabituated to the familiar stimulus and that the younger infants lookedmore to the familiar stimulus than the older infants. On the preferencetest trials, the 3- to 4-month-olds displayed a mean percentage preferencescore for the novel organization of stripes of 48.28% (SD 15.52), a valuethat was not reliably different from the chance preference of 50%, t(31) -0.04, p > .10, one-tailed. In contrast, the 6- to 7-month-olds showed amean novelty preference of 57.58% (SD 13.48), a score that wassignificantly above chance, t(31) 3.18, p < .005, one-tailed. The findings suggest that 6- to 7-month-olds, but not 3- to4-month-olds, can use form similarity to organize visual patterninformation. In combination with the outcome of Quinn et al. (1993), theresults imply that different Gestalt principles may become functional overdifferent time courses of development. Experiments in progress are aimed atproviding convergent evidence for this conclusion.
poster
We investigated whether young infants can orient reliably towards the“most salient” and the Ôleast salient' object in a visual scene, usingtexture-defined stimuli.We tested infants from 1 to 12 months of age in 4 age groups (1-3, 4-6,7-9 and 10-12 months old, n12 infants per group). Informed adults(n16), adults naive as to the purpose of the study (n16), and children3-4 years of age (n16) served as controls. The subjects were testedwith a forced-choice preferential looking technique.In a first series of experiments the subjects were tested with stimulipresented on two textured fields. In half of the trials one texturefield contained a group of larger blobs in a surround of small blobs(Ômost salient' task); in the other half, one texture field contained agroup of 16 smaller blobs in a surround of large blobs (Ôleast salient'task). Each session contained a minimum of 20 trials.All adult subjects showed high, statistically (p<0.01, one-sided Studentt-test) significant preferences for the discrepant group in both tasks.Preferences for the Ôleast salient' stimulus were slightly lower thanfor the Ômost salient' stimulus. For naïve subjects, performances inboth tasks were lower than for the instructed subjects. All groups ofinfants showed statistically significant preferences for the “mostsalient” stimulus. Children ages 3-4 years showed highly significantpreferences for both tasks, comparable to those of adult observers;infants under one year of age showed no preference for the Ôleastsalient' stimulus.These results were replicated in a second series of experiments, inwhich twelve 1-3 month-old infants and twelve 3-4 year-old children weretested with textures presented on cards. Control experiments indicatedthat this outcome was not due to the differential responses of younginfants to luminance increments or decrements.In an additional experiment, twelve 2-month-old infants were tested withslides in which the Ôleast salient' texture field was paired with atexture containing only small blobs. The infants oriented reliablytowards the texture containing the discrepant target, suggesting thatthe lack of preference for the Ôleast salient' target in the previousexperiments was not due to an inability to segment this target, butrather to a competition between the preference for a non-salientsegmented texture and a homogeneous texture consisting of salient items.These infants did not show a preference for the Ômost salient' texturecontaining the group of larger blobs, when paired with a homogeneoustexture containing only large blobs.Our results suggest that the attentional mechanisms responsible fororientation towards the Ôleast salient' object in a visual scene arefunctional in infants as young as 2 months of age. However, thispreference might be overshadowed by the preference for single, salientitems. It is the relative preponderence of the “forest” over the “trees”that is not functional in the young infant, but develops somewherebetween 1 and 3 years of age.
poster
Several studies have shown that the ability to perceive subjectivecontours develops sometime within the first 7 months of age (e.g.,Sireteanu, in press). Subjective contours are characterized by twofeatures: (a) they seem to be brighter than the surrounding surfacedespite the fact that there is no luminance-difference ('illusorybrightness') and (b) a form with a sharp border is perceived('illusory form'). The aim of the present investigation was to separate the two illusory phenomena from each other. Specifically, it was tested whether infants are capable of extracting the illusory form embedded in the Kanizsa figure, that is, whether illusory form can occur independent of illusory brightness. In order to eliminate illusory brightness from the Kanizsa contour, a stimulus configuration introduced by Prazdny (1983) was constructed. This stimulus consists of inducing elements with opposite contrast polarity, that is, of two white and two black 'pacmen', in front of a gray background. The idea here is that illusory brightness should be absent in this figure, since the inducers produce opposite brightness contrast effects which cancel each other out. Furthermore, since motion is an attractive feature for babies, the subjective contour or, rather, the illusory form was moving continuously back and forth. This was implemented by using a display composed of twelve pacman elements. By appropriately rotating the elements, one can make the illusory figure 'jump' across the screen. The first experiment was a habituation-dishabituation experiment. Four-month-old infants were tested for their capability of discriminating between an illusory Kanizsa square, which was continuously moving back and forth on a computer screen, and an illusory Kanizsa square, which was jumping through random positions. It should be noted that all pacmen were rotating all the time and that the amount of rotations within both animations were parallel meaning that the animations could be distinguished only through detecting the different movements of the illusory form. An ANOVA showed that the looking behavior of the subjects during the dishabituation phase was not determined by the foregoing habituation. Instead, there was a significant preference for the continuously moving subjective square. Taking up the results of the first experiment, a second study was designed as a natural preference investigation. Here, a further group of 4-month-olds was tested for their capability of distinguishing between the continuously moving illusory square and a likewise continuously moving symmetric figure with outward pointing cutoffs. Again, data analysis revealed a significant preference for the subjective contour. In order to exclude that this result was based on a preference for aligned elements, a control experiment was run. This experiment was an exact replication of the main condition with the exception that the pacman elements were given as outlines only, thereby destroying the illusory effect. In fact, no preference could be observed in this condition. To summarize, it can be concluded that 4-months-olds are able to perceive illusory form, if it is presented as a moving stimulus. These results are discussed within the context of more recent neurophysiological and psychophysical research (e.g., Curran, Braddick, Atkinson, Wattam-Bell, & Andrew, 1999).
poster
This presentation will describe a program of research that investigatesthe development of lightness constancy. In one experiment thedevelopment of the ability to distinguish between a dark region, causedby a cast shadow, and an equally dark region, created by a surface thatreflects little light was investigated in 5-and 7-month-old infants. Inthis experiment, infants were habituated to a cube whose left half waswhite and whose right half was black. A slide projector illuminated thecube; such the lower half was in deep shadow while the top was in brightlight. After the habituation criterion was met, infants were presentedwith two test displays on alternating trials. One test display consistedof the black and white cube with the shadow removed, that is, the entirecube was illuminated. The other test display presented the cube with theshadow unchanged but with the black area removed, that is, this testcube was all white but only its top half was illuminated. Controlsubjects were run to provide baseline measures of the attention-holdingproperties of the test displays. They older infants paid less attention,that is, looked for a shorter time, to a change in luminance of part ofthe display that was caused by a shift in the location of a cast shadowthan they paid to a shift in the location of a dark paint on the surfaceof the cube. In contrast the 5-month-old infants treated both shadowsand dark regions equivalently.In a second study, using displays based on the work of Adelson,3-year-olds and 7-month-old infants demonstrated a degree of lightnessconstancy while 5-month-olds did not. The infants were presented withan image of surface that appeared to be folded so parts of the surfacefaced the light source while the others were shaded. When the display isperceived as a folded surface, light and dark band are perceived thatrun perpendicular to the folds. These bands are not physically presentbut are the result of lightness constancy processes. Infants werehabituated to either vertical or horizontal bands and then tested withthe 'folded' display in both orientations. Seven-month-old infantslooked longer at the display with 'virtual' bands in a new orientation.The 7-month-olds responded to the reflectance of the surfaces and notthe intensity of the light that reached their eye. In contrast, the5-month-old infants did not.A third study is investigating the development of sensitivity to thelocation of a cast shadow. The location of the shadow cast by an objectcan indicate the location of that object. Preferential reaching to anapparently closer of two toys is being used to test 7-month-old infants.The infants tend to reach more often to the toy that casts a distantshadow, suggesting sensitivity of the depth cue.This work suggests that the visual world of the 5-month-old may bedifficult for the adult to imagine.
poster
It is a fundamental question whether newborn infants have an ability to perceive objects in a Gestalt manner or whether perceptual development require integration of elementary modules for features in a constructive manner. Previous studies provided evidence that newborn (Slater et al. 1991) and more than 3 months old (Bushnell & Roder 1985, Mundy 1985) infants may perceive objects as color-shape compounds. However, no one has studied perception of multiple objects with various features, where the binding problem may arise, and developmental changes in such perception in early infancy. We studied visual behaviors of 118 infants ranging in age from 12 days to 6 months using a method of infant-control habituation. Infants were shown simultaneous presentations of two objects which features consist of colors (red/green) and shapes (circular/triangular faces). The left/right positioning of the objects was changed between the trials. When a criterion for habituation had been met, only the combination of colors and shapes was changed for half of the infants (test group). Another half of the infants (control group) were offered two habituation trials additionally and then shown the display of the control trials. We observed changes in looking time of the last two habituation trials and the first two test trials. Significance of dishabituation was examined by comparison between the test and control groups. We found that the infants aged 0-1 and 3-6 months showed significant dishabituation to the new combinations of features, while only the 2-months-old infants showed no significant dishabituation due to spontaneous regression to familiar objects. Further video analysis of eye movements during these trials revealed that looking parameters, such as total looking time during the habituation trials, duration of fixation to one object and numbers of saccades per trial, showed inverted U shape as functions of age with clear peaks around at 2 months. This study indicates that the development of visual feature binding is not continuous. The mechanism of looking multiple objects with colors and shapes may change from the Geshtalt perception to the feature integration based on the modularity of the cerebral cortex at the end of 2 months.
poster
Basic visual functions such as object segregation are based on the processingof textural changes in the visual field. Adults discriminate textural changesbased on features more easily than those based on feature relations. Thissuggests that adults process individual features before processing theconjunction relations among these features. In the current study, we examinedthe nature of featural and conjunction processing in infancy. Experiment 1 investigated whether infants, like adults, exhibit a deficit inconjunction versus featural processing in visual textures. Using aninfant-controlled habituation procedure, 5.5-month-olds were habituated topatterns of colorful shapes and tested with a familiar pattern paired withanother that contained a discrepant patch. The discrepant micropatternswere ofnovel color (color condition), novel shape (shape condition), or novelrecombinations of familiar color and shape features (conjunction condition).Infants discriminated changes in both the color and shape features, but failedto discriminate changes based on conjunction relations between these features.The mean novelty preference to the discrepant patch was significantly greaterthat the chance level of 50% in both the color and shape conditions (colorcondition: 65.50%, t(11) 3.24, p < .01; shape condition: 57.52%, t(11) 2.11,p <.03), but not in the conjunction condition (48.79%, t(11) -.21, p >.05).These results indicate that 5.5-month-old infants failed to discriminatechanges in relations among features even though they encoded information aboutboth of the individual features that comprised these relations.A similarity-based model that does not assume differences between featural andrelational information processing could explain the above findings on thebasisof the greater similarity between the pattern containing the discrepant patchand the habituation pattern in the conjunction than in the feature conditions.We tested this account in Experiment 2 by exposing infants to patternscontaining conjunction discrepancies that were equally dissimilar to thefamiliar pattern as were those that contained feature discrepancies inExperiment 1. Similarity was measured by the overlap in the numbers offeaturesand conjunction relations between the new and the familiar micropatterns.Infants in this experiment still failed to discriminate the conjunctiondiscrepancies. Their mean novelty preference score of 50.04% was notsignificantly different from the chance level of 50% (t(11) .01, p > 1).Thus,overall similarity between the familiar and the novel pattern could notaccountfor the featural versus conjunction dissociation. These results argue against a similarity account of the dissociation ininfants' sensitivity to featural versus conjunction discrepancies, and suggestthat in infancy, as in adulthood, different processes are involved in thediscrimination of feature versus conjunction discrepancies in visual textures.
poster
To function effectively in this world, organisms need to be able to derive the 3-D structure and spatial layout of objects from the 2-D information that is available in visual images. Indeed, human adults readily derive 3-D information from static 2-D images. We examined the nature of this function in 3-month-olds. Specifically, we examined whether 3-month-olds are sensitive to line-junction information in static images that adults use to derive the 3-D structure of objects, and whether discrepancies in such information 'pop-out' and engage infants' attention.Prior research has demonstrated that, akin to the effects of pop-out in adults, discrepancies based on fundamental features, such as line-crossings and orientation, attract and engage infants' attention. Prior research also suggests that discrepancies in images that appear to have 3-D structure pop-out for adults. In the present study, we examined whether discrepancies in 3-D cues also engage infants' attention in the same manner as discrepancies in simple fundamental features such as line orientation.Using an infant-controlled habituation procedure, 3-month-olds were habituated to 4 X 4 arrays of line-drawings that appeared to adults to be blocks oriented in a particular direction (3-D condition) or to comparable images that appeared to adults to be flat 2-D line-drawings (2-D condition). Following habituation, infants were tested for their preference between two patterns, one pattern containing a single block oriented in a novel direction among 15 familiar blocks and another containing a single familiar block amidst 15 novel blocks. In other words, infants were tested for their preference between an array that had a single novel micropattern and a majority of micropatterns that were familiar versus another array that had a single familiar micropattern and a majority of micropatterns that were novel.Infants in the 3-D condition preferred the array with the single novel element (even though the majority of elements in this array had a familiar orientation); in contrast, infants in the 2-D condition preferred the opposite array. The mean preference score for the side with the individual misoriented element for infants in the 3-D group (58.76%) was significantly greater than the chance level of 50% in the 3-D condition, t(17) 2.44, p <.03, whereas this preference was significantly less than 50% in the 2-D condition (42.48%), t(17) -2.26, p < .04. Thus, performance in the 3-D condition was determined by the novelty/familiarity of the individual discrepant objects in the test arrays, whereas performance in the 2-D condition was determined by the novelty/familiarity of the majority of elements that surrounded the discrepant elements. In other words, the individual novel discrepant element in one of the test patterns appears to have engaged the infants' attention in the 3-D condition but not in the 2-D condition.These results suggest that, akin to pop-out in adults, discrepancies in 3-D cues derived from static images engage 3-month-olds' attention. Thus, early in life, humans selectively attend to ecologically relevant 3-D cues in static images.
poster
In illusory contour displays, adults typically perceive both a centralillusory figure and a set of complete, partly visible inducingelements. In a Kanizsa illusory square, for example, four incompletecircles with aligned, missing quadrants are perceived as an illusorysquare resting atop four complete circles. Recent theories of illusoryfigure perception have highlighted the importance of the amodalcompletion of inducing elements for perception of the illusion (Kellman& Shipley, 1990), but previous studies of infants have not investigatedthe development of this aspect of the illusion. The present researchinvestigates whether infants perceive illusory contours by examiningwhether they too perceive the amodal completion of inducing elements. Positive evidence for amodal completion would offer the strongestevidence that infants organize illusory contour displays as do adults.In Experiment 1, 7-month-old infants were habituated to either aKanizsa illusory square or a control display composed of the sameinducing elements, rotated so as to form a non-illusory pattern. Following habituation, infants were tested with a single inducingelement (a circle with a missing quadrant) and a single completecircle. Results showed that following habituation to the illusorysquare, infants dishabituated to a single inducing element, despite itssuperficial familiarity, and transferred habituation to a completecircle. In contrast, infants habituated to the control non-illusoryfigure dishabituated to the complete circle, demonstrating that theycould distinguish the test displays, recognized the elements in thenonillusory habituation display, and had no inherent preference for theincomplete inducing element. These results therefore provide evidencethat 7 month olds perceive illusory figures against a background ofamodally completed elements.In Experiment 2, 4-month-old infants were tested with the samedisplays. Results showed that 4 month olds looked equally at thecircle and the inducing element test displays following habituation toboth the illusory and non-illusory control displays. The lack ofpreferences in the control condition suggests that the younger infantswere unable to discriminate the two test displays; a failure thatprecludes any strong interpretation of performance in the experimentalcondition. In Experiment 3, 4-month-old infants were habituated to a larger, moresalient illusory figure than the illusion presented in Experiment 2,while all other aspects of the design remained the same. In contrastto both previous experiments, infants who habituated to this strongerillusory figure dishabituated to the complete circle, providingevidence they failed to perceive the illusory figure and interpretedthe illusion display as a collection of distinct inducing elements.This research provides compelling evidence that 7-month-old infantsperceive illusory figures in the same way as adults, whereas4-month-old infants do not. These findings are in accord with previousresearch on the development of perception of illusory figures(Bertenthal, Campos, & Haith, 1980) and of depth (Yonas & Granrud,1985). They suggest that a qualitative change in perceptualorganization occurs near the middle of the first year of life.
poster
We perceive a world composed of coherent, segregated objects, despite theabsence of visible boundaries of partially occluded objects. In addition,as objects move we perceive them as persisting even when temporarilyoccluded by other objects. These abilities to fill in the gaps inperceptual experience are fundamental to human perception, and vitalquestions exist regarding their developmental origins.By 4 months of age, infants appear to perceive the unity of two rod partsextending from behind an occluder (Johnson & Aslin, 1996; Kellman & Spelke,1983). Perception of object unity involves interpolating object parts, orfilling in a spatial gap. In like fashion, when an object moves behind anoccluder, the observer must interpolate its path, or fill in aspatiotemporal gap. Such events have been presented to infants, and theresults interpreted in terms of infants' reasoning and knowledge of objectidentity (e.g., Baillargeon, 1986; Spelke, Breinlinger, Macomber, &Jacobson, 1992). These interpretations are, however, a focus ofcontroversy, it being argued that lower order perceptual factors canexplain the results (Bogartz, Shinskey, & Speaker, 1997; Haith, 1998).Our motivation for the present studies is the fact that little has beendone to systematically explore infants' perception of object trajectories.We began our investigations by presenting 10 4-month-olds a ball repeatedlymoving laterally behind an occluder, until habituation. The ball moved ata rate of 7.2 deg/s, was completely occluded for .73 s, and was visible inits entirety at either side of the occluder for 1.07 s. After habituation,the infants viewed a broken trajectory, in which the occluder was replacedby a gap (i.e., the ball seemed to disappear at an edge and then reappear),and a complete trajectory, in which the ball was visible during the entireextent of motion. A control group of 10 infants viewed the broken andcomplete trajectories with no prior habituation. The experimental grouplooked significantly longer at the complete trajectory (M 21.43 s, SEM 4.22) than at the broken trajectory (M 13.15 s, SEM 3.12), t(9) 2.70,p < .05. The control group did not exhibit a reliable preference (Mcomplete 22.44 s, SEM 3.89; M broken 27.72 s, SEM 3.77), t(9) 1.11, ns. The difference in performance across groups was significant,F(1, 18) 5.77, p < .05. A third group of 10 infants viewed a ball movingbehind a narrow occluder, completely occluded for .07 s. During test,these infants looked significantly longer at a broken trajectory (M 11.75s, SEM 3.26) than at a complete trajectory (M 7.01 s, SEM 3.27), t(9) 4.61, p < .01. The difference in performance relative to thewide-occluder condition was significant, F(1, 18) 16.22, p < .001.These results suggest that the infants in the wide-screen condition failedto fill in the spatiotemporal gap in the ball's trajectory. Infants in thenarrow-screen condition, in contrast, seemed to perceive the continunity ofthe trajectory. These findings indicate that young infants'representations of occluded objects may be transient except under verybrief occlusion. This claim appears to conflict with a view of younginfants as capable of reasoning about object properties under occlusion.Research is underway to probe more fully the reasons underlying thesediscrepancies.
poster
no abstract