非表示:
キーワード:
-
要旨:
In visual object recognition, it is important to understand which object properties are important for learning. Typically, this is done by comparing recognition performance across experimental conditions that manipulate and isolate different aspects of object properties e.g., distinctive features. However, such an approach requires object properties to be explicitly specified prior to testing and is, hence, limited by the experimenter’simagination (or the lack thereof). Here, I will present a different approach to studying this problem. Rather than predefine the object properties of interest, participants are free to explore all aspects of a set of novel 3D objects during learning. Raw data are collected on observers’ patterns of exploration and analyses are subsequently applied to understand which object properties are valued by the observers during learning. In my presentation, I will describe the technical apparatus that supports this experimental approach. In addition, I will provide details on how raw data are collected and the methods of post-hoc analyses that can be applied to the data.
There are several advantages to this approach in addition to those already mentioned. Firstly, this approach places control in the hands of the observer. Thus, stimulus presentation is determined by the observer’s goals rather than the experimenter’s preconceptions. This results in findings that are closer to ecological validity. Also, the raw data lend itself to reanalysis when new methods of analyses are devised or when previously unconsidered object properties later prove to be relevant for object learning.
The purpose of this presentation is to generate an open discussion on the merits and disadvantages of this approach to studying visual object recognition.