CHAPTER 4 RESEARCH QUESTION AND STUDY APPROACH
4.3 Implementation of the Research Question
In order to elucidate the premise of this hypothesis a task was generated that allowed testing the influence of motion information on object individuation. The violation-of-expectation paradigm proved to be useful to determine whether infants are able to individuate objects (Baillargeon, 1998; Wilcox & Baillargeon, 1998b; Xu & Carey, 1996). According to Xu and Carey (1996), their event-mapping task admits the investigation of the role property/ kind information plays in infants’ object individuation capacity. Closely following the procedure described by the authors (cf. Xu & Carey, 1996, Experiment 2), infants are first familiarized with an event in which two different objects successively emerge from behind either side of a screen. During test trials, infants are presented with displays that comprise either one object (unexpected outcome) or both objects (expected outcome). Differing from the original
study where objects showed uniform motion21, the task to be used in the present context displays domain-specific motion patterns. Motion contains various types of information, which can be divided into path and manner information. Path information refers to the route an object takes when moving (i.e. from behind an occluder to the left side of a stage) such as depicted in the Xu and Carey experiments. Manner information on the contrary describes the way the object moves (e.g., Choi & Bowerman, 1991; Choi et al., 1999). Whereas path of motion provides location information, it is more likely that manner of motion can be used as an identifying feature of the object itself. In the current work manner information is made salient to activate conceptual knowledge about animates and inanimates in order to serves as kind information in the individuation task.
To find out whether infants deduce certain expectations from the additional kind information, revealed during the familiarization trials, looking times toward the outcome scenes are checked against initial looking preferences22 at one-object versus two-object displays with the same objects used in test. Only if this comparison shows a significant shift in the expected direction (i.e. longer looking toward one object after familiarization) can one infer object individuation on the basis of property/ kind information that distinguished the objects. Xu and Carey (1996) as well as researchers who modeled experiments after theirs (e.g. Baillargeon & Wilcox, 1998b, Experiment 1 and 2; Bonatti et al., 2002; Surian et al., 2004) implemented this in a between subject-design. That is, they used separate groups when comparing the preference for looking between baseline and test trials. Unlike previous work, the present studies applied a within-subject design meaning the outcome displays are presented as baseline before the start of the familiarization phase within one single session. This change in method seemed necessary because some experiments detected a priori looking preferences for the two-object display whereas others did not (e.g., Surian et al., 2004). Using the same stimuli for baseline and test displays (and for the same subjects) guarantees that significant results supporting the initial hypothesis cannot be attributed to either differences in stimuli or a-priori group differences. Pauen and Träuble (2004) successfully worked with this within-subject testing method to
21 During familiarization the objects in Xu and Carey’s (1996) study were moved back and fourth on slightly visible sticks attached to the bottom of each object.
22 In order to acquire a possible initial preference infants simply view the outcomes of the test trials without being exposed to any familiarization emergences before. This is referred to as baseline condition or baseline phase.
investigate knowledge-based reasoning. Their task was arranged in three separate scenes: A baseline scene (1) which was identical to a test scene (3) that examined the impact of a motion scene presented in second place (2). Testing 7-month-old infants, Pauen and Träuble provided evidence for the use of previously acquired knowledge about causal behavior of animals and inanimate objects guiding infants’ interpretation of an ambiguous sequence of moving objects. The experiments to be described next combine aspects of the Xu and Carey procedure with the Pauen and Träuble task. More specifically, infants’ preference for either one or two object displays was tested by comparing looking times during a baseline phase, which took place before the familiarization scenes with an occluder and a test phase that followed familiarization. Problematic with this approach is, however, that spatiotemporal information for two objects is given when presenting the two-object display which might result in a confounding of information available for infants to rely on. One might argue that the information provided by a within-subject baseline influences infants’ reaction in test. Despite that possibility this type of design was preferred according to the reasons explained above. Due to the within-subject design in which a baseline is presented before familiarization the following experiments forego additional introductory trials.
In terms of the experimental presentation, Xu and Carey (1996) demonstrated the occlusion events live in a puppet stage and used an infant-controlled design. For the following set of studies a video presentation and a fixed-trial procedure23 was chosen. Several reasons speak for this line of action: The first is to assure maximal standardization in that each infant views the same animate movement pattern. It would be extremely difficult to avoid variations and irregularities in a life display involving animate motion. Another reason for a film presentation is the reduction of spatiotemporal information available to influence infants’ interpretation regarding the number of objects involved in the event. So if infants show a preference for one over two objects, it can be attributed to infants’ knowledge about the kind of visual entities performing the movements. Thus, no additional information besides global level category membership (expressed by characteristic motion information) is supposed to be present. In addition, results offered by Seekircher (2007) support the assumption that infants of the tested age-range are able to extract information from a film presentation equally well as from a live demonstration. When demonstrated per video
presentation how a certain part of an object functions, 12-month-old infants made use of the information illustrated in the movie. In a subsequent live completed categorization task with similar three-dimensional objects consisting of various functional partitions infants discriminated in accordance with the critical part shown in the films. The data of this study was comparable to results obtained by Träuble and Pauen (2006) who worked with the same material and a similar procedure with the only difference being the demonstration of the function of the critical part, which took place live (Seekircher, 2007). Not only did infants extract crucial pieces of information from video but also did they transfer newly acquired knowledge to a live task. Other studies conducted by Madole and Cohen (1995), Perone and Oakes (2006), as well as Mumme and Fernald (2003), showed that infants gather, process, and apply information from videos in a variety of contexts. Surian et al. (2004) further demonstrated that infants performed successful object individuation by utilizing a video technique. Based on these findings, it seems well justified to presume that infants of the tested age-range process information in similar ways when presented in the format of live and video displays.
The following chapters explain the design, the stimuli, the experimental setting, and the procedure of the experiments in more detail.
CHAPTER 5
METHODOLOGY AND PARADIGM