Theory
Event categorization is a framework on how infants form and use representations of physical events. Baillargeon, Wilcox, Needham and their colleagues see individuation as one specific problem of infant’s event categorization across different physical domains (e.g. Baillargeon & Wang, 2002; Needham & Baillargeon, 2000; Wilcox & Schweinle, 2002). Investigators holding this view assume that infants assign physical situations to broad categories including occlusion, support, arrested-motion, and containment and build up a mental representation of the physical event watched with respect to spatial, temporal, and mechanical information (Baillargeon, 1998; Leslie, 1994; Xu and Carey, 1996). Thus, not only do infants have to categorize the available information into a simple structure that makes up the event. In addition, they have to consult already stored information about the specific kind of event and confine it from other kinds of events (Arguiar & Baillargeon, 2002; Baillargeon & Wang, 2002; Hespos & Baillargeon, 2001; Wilcox & Schweinle, 2002). According to the model of infants’ acquisition of physical knowledge (Baillargeon, 1994, 1995, 1998, 2000; see Figure 2), a specialized learning mechanism is held accountable for the formation of physical categories, which
correspond to distinct ways in which objects behave and interact. When learning about a physical category, infants first establish a “preliminary, all-or-none concept” that captures only the core of the category (Baillargeon, 1994). Typically this includes only basic spatial and temporal information as specified in principles of continuity and solidity (Baillargeon, 2004). During the course of development the initial concept is progressively elaborated and refined. With further experience, infants identify more variables that are relevant to an event category11 and incorporate this additional knowledge into their reasoning. As a result infants are able to make increasingly accurate predictions and interpretations over time (Baillargeon, 1999).
Figure 2. The physical reasoning account. From Baillargeon (2004).
For instance with respect to object individuation, Baillargeon and her colleagues provide evidence that precursors of this competence are found in infants as young as 2.5 months. However, even though infants at this age expect that an object continues to exist after it becomes hidden (Aguia & Baillargeon, 1999; Spelke et al., 1992; Wilcox et al., 1996), their knowledge about occlusion events is still incomplete and it takes the identification of relevant variables such as shape or color to improve the ability to individuate. Therefore, the acquisition of knowledge about occlusion events follows the same developmental trend as observed in other physical categories showing that the range of violations, infants solve, increase with age (Baillargeon,
11 Learning occurs separately for each event category since these variables are not transferred between relevant categories (Baillargeon, 2004).
1999). Whereas infants at 3.5 months of age identify height as an occlusion variable it is not until 7.5 months of age that infants include transparency in their judgments (Baillargeon & DeVos, 1991; Luo & Baillargeon, 1994).
From the event categorization perspective, occlusion events confront infants with a special problem. In order to be able to individuate the objects which are part of an occlusion event, infants not only must decide whether the entities successively seen on either side of an occluder constitute one or two distinct objects, they also have to determine if the sequence they view consists of one or more events (Baillargeon, 2004, Wilcox & Baillargeon, 1998). Wilcox and Baillargeon (1998) contended that in Xu and Carey (1996), the infants were confronted with a task that involved two separate categories of events. Whereas during familiarization they saw an occlusion event in which objects moved back and forth behind a screen, they viewed a non- occlusion event in which objects rested on a platform in test with no screen being present. According to the event categorization approach, the presence or absence of the screen creates a crucial difference between familiarization and test events and produces a change in event category. As a consequence, instead of viewing the screen’s removal as a change in an ongoing situation, infants may view it as the start of a separate physical situation. Such being the case infants reclassify and initiate a new event representation. Further, when presented with two different physical situations infants not only have to establish two separate event representations, they also must form a link between them (Baillargeon, 2004). This linking or event- mapping requires several processing steps. In order to follow and make sense of categorically distinct situations infants need to remember what occurred in one event, map this information onto the ongoing in a second one and compare the two events. The more complex an event sequence, the more difficult this process becomes. Xu and Carey’s event-mapping task involved featurally distinct objects that follow complicated trajectories. Thus, infants had to judge whether the objects’ movements and interaction are consistent with their existing knowledge (Wilcox et al., 2003). The familiarization event is supposed to build up an expectation that is tested in the events during test. This can only be done if the two events are perceived as belonging together. Proponents of the event-categorization approach doubt that infants perceive a single continuous situation in event-mapping tasks on which the ‘Object-first- Hypothesis’ is based on. Thus, the failure to respond correctly in an event-mapping task at 10 months of age could be attributed to the inability to complete the mapping
process during test. That is, infants were not able to judge whether the event seen during familiarization mapped onto the one seen during the test phase. So it could be that infants successfully individuate objects at this age, but they were simply not able to reveal this ability within the context of an event-mapping task. Therefore, event- mapping is seen as a limitation that results from infants’ bias to form distinct physical categories and to reason and learn in terms of these separate categories (Wilcox and Baillargeon, 1998). It alters infants’ categorization of the physical situation presented to them. What speaks for this hypothesis are results maintained by tasks explained in the proceeding paragraphs.
Method: Event-monitoring tasks - object changes
Evidence for this presumption that infants under the age of 12 months are able to individuate objects comes from studies concerned with event categorization showing that infants group physical events into different categories (Baillargeon, 1995, 1998; Hespos and Baillargeon, 2001). Based on the assumption that the main difficulty with event-mapping is the retrieval of a clear representation of the occlusion event, Wilcox and Baillargeon’s (1998b) designed a new procedure, namely event- monitoring task. In order to aid infants in accessing an event representation they showed a continuous occlusion event in which infants see only one event involving one or two objects. By doing this, infants do not need to engage in event mapping and thus do not have to compare an earlier familiarization or habituation event with a test event but focus on the test event itself (Wilcox & Baillargeon, 1998b). Therefore, the only task infants have to solve is monitoring whether the event is consistent in itself. This makes event-monitoring tasks less cognitive demanding in comparison to event- mapping tasks, because event-monitoring involves only one processing step whereas event-mapping incorporates multiple ones. Empirical findings suggest that it is presumably easier for infants to monitor the internal consistency of one event compared to the mapping of one event representation to another (Aguiar & Baillargeon, 2002; Hespos & Baillargeon, 2001; Leslie et al., 1998; Wilcox, 1999; Wilcox & Baillargeon, 1998b; Wilcox & Schweinle, 2002). In a large number of studies that applied an event-monitoring task the ability to use featural information to individuate objects was demonstrated with infants aged 4.5 to 11.5 months (e.g. Wilcox, 1999; Wilcox & Baillargeon, 1998a, b; Wilcox & Chapa, 2002). Three
general approaches can be divided: (1) “the single-trajectory experiments,” (2) “the narrow/ wide screen experiments,” and (3) “the opaque/ transparent occluder experiments”.
The single-trajectory experiments
In the single-trajectory experiments Wilcox & Baillargeon, (1998b, Experiments 8) used a simplified version of the Xu and Carey (1996) task. For one thing this was done by making the events shorter. During familiarization the objects just moved from left to right without reversing their trajectory (Wilcox & Baillargeon, 1998b). For another thing this was accomplished by using only one test display that contained a single object. Instead of checking looking times to a one-object test display against a two-object test display, they compared infants’ reactions to a one- object test outcome. Therefore, they randomly assigned infants to one of two conditions: the box-ball condition or the ball-ball condition. In the box-ball condition, infants were familiarized to a sequence of a box moving from one side of the stage and disappearing behind a screen, followed by a ball emerging from the other side. In the ball-ball condition on the other hand, infants saw a ball going behind an occluder and the same ball coming out the other side. The screen was then lowered to reveal a single ball on the stage in both conditions. Infants looked longer at the single ball outcome in the box-ball condition than in the ball-ball condition. Wilcox and Baillargeon (1998b) concluded that the infants must have used perceptual property information to establish a representation of two distinct objects. Therefore, the single ball outcome was unexpected in the box-ball condition. Thus, the authors were able to show that infants at 9.5 months of age individuate objects by means of object properties alone when the number of object trajectories involved in the introduction/ familiarization sequence was lessened and the experimental procedure was simplified. Later work by Wilcox & Schweinle (2002) suggests that the age in which the ability to individuate object could be shown reduced to 5.5 and 7.5 months in the case of just a single trajectory. Thus, when the task is sufficiently simple, infants younger than 10 months of age appear to be able to individuate objects on the basis of their perceptual features. Wilcox et al. (2003) present evidence that young infants are able to use featural information as the basis for object individuation.
The opaque/ transparent occluder experiments
Wilcox and Chapa (2002) proposed another way of tailoring the procedure to the information processing capacity of younger infants. In their version of the event- monitoring paradigm they made the task simpler and more traceable by employing a transparent occluder in the test events. Thus, 9.5-month-old infants viewed either one object (i.e. a ball) or two objects (i.e. a box and a ball) emerge successively to opposite sides of an opaque occluder. When the screen was lowered a single ball behind a transparent screen was revealed. This was compared to a condition without an apparent screen standing behind the occluder. Only the infants who saw the ball in the transparent screen condition correctly judged that the one-ball display was inconsistent with the box–ball sequence. The authors’ interpretation of the results was that infants categorize events involving opaque and transparent occluders as the same kind of physical situation (i.e. occlusion). Thus, infants only had to engage in event- monitoring of a single event, which according to Wilcox and Chapa (2002) enabled them to solve the object individuation task. For this reason the findings support the notion that infants are more likely to give evidence of object individuation when they need to reason about one kind of event (i.e. occlusion) than when they must retrieve and compare categorically distinct events (i.e. occlusion and no-occlusion).
The narrow/ wide screen experiments
Besides simplifying the task demands Wilcox and Baillargeon (1998b) thought of an additional way to assess infants’ interpretation of occlusion situations with an event-monitoring task. They presented 9.5-month-old infants with a sequence in which a red ball disappeared behind a screen and after a brief interval a blue box emerged at the other side of the occluder. Subsequently, the box reversed its trajectory and vanished behind the screen followed by the ball appearing on the other side. This event was presented without interruption for the time the infants kept their attention on the stage area. In order to keep a continuous event even during test the screen was not removed to reveal one or two objects. Instead of lowering the occluder in test trials, infants had to judge whether the screen was sufficiently wide to hide the two objects simultaneously (Wilcox & Baillargeon, 1998b). Two conditions were checked against each other: a narrow-screen condition in which the screen was too small to fit
both objects behind side-by-side and a wide-screen condition in which the screen was broad enough to fit both objects simultaneously (Xu, 2003). The hypothesis was that if infants were led by the perceptual property differences between the ball and the box to conclude that there were two distinct objects, they would look longer at the narrow- screen event because the two objects could not fit behind the small screen at the same time. Wilcox and Baillargeon (1998b) obtained this result and concluded that when the experimental task was modified in this way, 9.5-month-old infants were able to use perceptual property/ featural information for object individuation. This finding could be extended to younger infants of 7.5 and 4.5 months (Wilcox & Baillargeon, 1998a, b). Using the same methodology, Wilcox (1999) investigated the features (shape, size, pattern, and color) infants were sensitive to. Infants 4.5 to 11.5 months of age were tested on displays in which the objects differed only in one perceptual property (e.g., size or color) at a time. Results indicate a developmental trajectory: Infants 4.5 months of age looked longer at the narrow-screen event when shape or size alone changed, but they did not look longer when surface pattern or color were manipulated solely. At 7.5 months, infants used the change in surface pattern to reason about the number of objects involved in an occlusion event and it was not until 11.5 months that infants included the color change in their judgment. Wilcox (1999) interpreted these results as evidence that infants at various ages use different types of perceptual properties for object individuation.
Three examples of event-monitoring tasks provided evidence that infants, much younger than 10 months of age, are capable of individuating occluded objects by means of featural information. Thereby these studies show that when infants must rely on property information as opposed to spatiotemporal information, event- mapping tasks in which infants are asked to relate an occlusion event with a no- occlusion event are more challenging for infants than event-monitoring tasks in which they have to reason about only an occlusion situation (Aguiar & Baillargeon, 1999; Baillargeon & DeVos, 1991). The more over, these experiments yielded detailed information about the timetable and hierarchy for the development of this competence. When given an occlusion event in which infants can only draw on featural information to individuate the objects in an occlusion situation they succeed in case the same objects are involved at some point between the ages of 2.5 to 10 months (Baillargeon, 2004a; Spelke et al., 1995; Xu & Carey, 1996). In occlusion
events in which different objects are involved infants succeed between 10 and 12 months of age (Xu & Carey, 1996). Despite her original claims, Xu has meanwhile acknowledged that younger infants are able to successfully include featural information in the process of individuation under certain conditions such as a simplified experimental procedure (Xu, 1997, 1999, 2002, 2003, 2007). Nonetheless, the author maintains the conception that spatiotemporal information is primary compared to featural information and that older infants make use of more references for individuation. Her advanced account on the development of object individuation is delineated next.