Symposium
Chair: Neil Berthier
Discussant: Marshall M. Haith
Recent work has suggested that young infants have considerable physicalknowledge about events and that infants are able to reason about suchevents and develop expectations regarding the state of the world (e.g.,Spelke, et al., 1992). These abilities have been demonstrated in experimentswhere infants evidence such abilities by looking longer at physicallyimpossible events. A major open question at the current time is whetherthis knowledge and ability to reason is limited to preferential lookingor whether this knowledge is of a general form and can be used toguide arbitrary action. The generation of action requires real-timeattentional control to the relevant aspects of a task, the storageof relevant information in working memory, the generation of actionplans, and the coordination of multiple action systems such as lookingand reaching. The current symposium will investigate these aspectsof coordinating representation with action and discuss whether thephysical reasoning abilities demonstrated in preferential lookingexperiments can be used by infants in situations where such knowledgemust be used to generate actions.The first presenter will discuss results from experiments with 7- and9-month-old infants in a task where the infants must prospectivelyreach or visually track an object moving in and out of occlusion.In those experiments the critical test trials involve a barrier placedbehind the occluder that stops the rolling object. The presenter willdiscuss under what circumstances infants were sensitive to barrierand how these data relate to neural systems that relate objects. Thesecond presenter will discuss studies of from 7- and 9-month-oldswere measures of intentionality provide important information aboutthe nature of representations that regulate infant manual search.The third and fourth presenters will discuss experiments with 2- to3-year-olds where children were required to search for a rolling objectthat was stopped by a partially observable barrier behind an occluder.Their results show surprising shortcomings in the abilities of thechildren to search for the ball effectively. These speakers will discussthe complexity of generating action in a search context, the difficultyin coordinating actions, the role of working memory, and the difficultyof inhibiting perservative actions in explaining the general failureof their children in solving these search tasks.
Details of individual items:
paper
The origins and early development of infants' physical knowledge aboutthe world is a source of intense debate. Contrary to the traditionalview expoused by Piaget and others that infants younger than 7 to9 months of age do not appreciate the continued existence of objects,an impressive number of recent studies suggest that infants show sensitivityto the continuity, solidity, and coherence of objects from the firstfew months after birth. One distinguishing feature of this recentresearch is that it relies on preferential looking, which demandsminimal coordination between object knowledge and actions.In order to better understand the rate limiting factors in the coordinationof object knowledge and actions, we conduceted a series of experimentstesting infants' abilities to use this knowledge in the control oftheir predictive tracking and reaching. The stimulus event was similarto the one used by Spelke et al. (1992) in which infants are showna ball rolling across a stage behind a screen and then reappearingon the other side. In our task, infants were tested for predictivetracking and reaching as the ball reappeared from behind the screen.On alternating test trials, infants were shown, prior to raising theoccluding screen, (1) a wall placed directly behind the location ofthe screen or (2) no wall behind the screen. If infants are capableof coordinating the requisite object knowledge with their trackingand reaching, then their predictive actions should be disrupted bythe presence of the occluded barrier.The first experiment tested predictive tracking and reaching with 9-month-oldinfants. On amost 50% of the trials, infants anticipated the reappearanceof the ball and their reaches were kinematically identical on Walland No-Wall trials, Yet, a subset of infants aborted their reacheson Wall trials suggesting that they appreciated the solidity and continuityof hidden objects and coordinated this knowledge to control theirpredictive reaching. Surprisingly, this disruption of reaching wasnot paralleled by disruption of tracking suggesting that the latterresponse is either independent of object knowledge or controlled byother factors when coupled with reaching. The second and third experiments tested these alternative interpretationsby modifying the experimental paradigm so that 9-month-old infantscould track the rolling ball but could not reach for it. Infants showeda disruption of predictive tracking when the wall was present (Experiment2), but this disruption occurred only when the barrier was placedin the path of the ball (Experiment 3). Finally, the results fromExperiment 4 revealed that some visual reminder of the barrier isnecessary to produce sufficient activation of object knowledge tointerfere with object tracking.Taken together, these results suggest that 9-month-old infants possesssome degree of object knowledge which is sufficient to guide theiractions sometimes. It is important to emphasize, however, that thisknowledge and its coordination with action is fragile and varies asa function of task and response measures.
paper
Use of problem-solving tasks to reveal representational abilities ininfant search Despite over 30 years of research into the developmentof manual search, two issues concerning infants' representations ofhidden objects remain controversial. The first involves the fact that,although young infants can construct and access representations ofhidden objects, manual search is delayed until 8 months. Munakataet al. (1997) argued that the late onset of manual search may be explainedby the development of representations of hidden objects that graduallyachieve the capacity to control manual behavior, rather than the developmentof appropriate means-end ability. They tested 7-month-olds on object-retrievaltasks for which it was claimed they possessed means-end ability. Infants'success at retrieving an object was markedly poorer when it was hiddenthan when it was visible. However, it is uncertain whether infants in this study demonstratedmeans-end ability. The principal measure was object retrieval, butrecent work has shown that infants who fail to produce means-end sequencescan also 'solve' means-end problems. Instead, measures of intentionalityare needed to identify whether infants use means-end sequences tosolve problems. A new study is reported in which 7-month-olds weretested on tasks in which a screened object could be retrieved by pullinga cloth. Previous work has shown that infants first produce means-endsequences on support tasks at age 7 months. Infants produced moreintentional behavior on tasks where a toy could be retrieved comparedwith control tasks where there was no toy, and maintained this differenceregardless of whether the toy was covered by either an opaque or transparentscreen. These findings suggest that young infants' representationsof hidden objects are sufficient to regulate manual behavior, andthat the relatively late onset of manual search is related to developmentof means-end behavior. The second issue involves infants' representation of the location ofa hidden object when making A not-B search errors. Diamond has arguedthat perseverative errors may be explained by poor inhibitory controlof reaching, rather than failure to maintain a representation of thecorrect location of a hidden object. Reports that infants displaydifferential visual attention to places A and B are consistent withthis interpretation. However, these findings merely demonstrate thatinfants distinguish between A and B, but do not indicate whether theycorrectly represent the location of the hidden object. Infants' representation of the location of the object was examinedin a study of 9-month-olds who first searched for an object at A.On B trials the object was secretly removed from the apparatus afterit was hidden. Thus, infants searching at A or B found no object ateither place. Infants' behavior was more intentional when searchingat B than at A. Additionally, infants engaged in other behaviors suchas looking on the floor after finding no object at B, but never producedsuch behavior after searching at A. These results suggest that searchon A not-B is driven by an enduring representation of the correctlocation of the object that can organize a sequence of intelligentsearch behaviors. These findings show that measures of infant problem solving, particularlymeasures of intentionality, can provide important information aboutthe nature of representations that regulate infant manual search.
paper
Recent work using the preferential looking method suggests that younginfants have considerable physical knowledge about events. These resultscontrast strongly with earlier work suggesting that young infantsare limited in their use of mental representations and reasoning inmanual search tasks. Action may require the infant to inhibit an incorrectresponse, to integrate physical knowledge with action planning, andto have precise knowledge about the location and orientation of objects.We have performed a pair of studies that investigates whether thereasoning abilities demonstrated by preferential looking studies extendto a manual action task in which inferences from physical knowledgemust be made.In our experiments, 2-, 2.5-, and 3-year-old children were requiredto open one of four doors in an occluder to reveal a toy ball. Thelocation of the ball was determined by an event the child viewed atthe beginning of the trial. Trials were initiated by placing a barrieron the track in one of four locations corresponding to one of thefour doors. The ball was then rolled behind the occluder and was stoppedby the wall. The child was then allowed to open one of the doors toreveal the ball. The results showed that only the oldest group ofchildren were able to reliably choose the correct door.The second experiment tested 2.5-year-olds in a task where the saliencyof the barrier wall was increased. Performance of the children wassignificantly better than in the first study, with girls showing superiorperformance to boys. In contrast to the first study, where the 3-year-oldswho solved the task appeared to use physical reasoning, children inthe second study seemed to solve the task on an associative basis. The results as a whole suggest that while young children are sophisticatedin their employment of sensorimotor skills, that the coordinationof sensorimotor behavior with reasoning and representational knowledgeis limited. The presentation will conclude with a discussion of howspatial reasoning and working memory are related to sensorimotor action.
paper
Two-year-olds' knowledge of solidity was tested in four search tasksadapted from infant looking time experiments of Spelke at al. (1992).In Experiment 1, 2-year-olds failed to reason correctly about thelocation of a falling ball after a hidden shelf that blocked its trajectoryhad been inserted in the apparatus. Experiment 2 extended this findingby demonstrating that both the effects of removing and inserting ashelf were not fully understood. Experiment 3 examined sensitivityto the constraint provided by a solid barrier on horizontal motion.In all three experiments, 2-year-old children searched initially atthe location where they saw the object during familiarization. Experiment4, using multiple test trials but no familiarization to a pretestlocation, also demonstrated the failure of 2-year-olds to reason aboutsolidity and support constraints on the trajectories of falling objects.Experiments 1 and 3 also included 2 1/2 -year-olds, who succeededon these search tasks. The implications of the poor performance of2-year-olds, in the face of success by very young infants on lookingtime measures of sensitivity to similar constraints on object motion,are discussed. The toddlers' failure may be attributable to a numberof factors including the need for prediction in these tasks but notin the looking time versions and a greater sensitivity of action planningto perseveration.