The Impact of Category Type and Working Memory Span on Attentional Learning in Categorization - eScholarship

The Impact of Category Type and Working Memory Span on Attentional Learning in Categorization Mark R. Blair (mblair@sfu.ca) 1 Lihan Chen (lca28@sfu.ca) 1 Kimberly M. Meier (kmm1@sfu.ca) 1 Michael J. Wood (mjw6@sfu.ca) 1 Marcus R. Watson (marcusw@psych.ubc.ca) 2 Ulric Wong (uwa@sfu.ca) 1 Cognitive Science Program & Department of Psychology, Simon Fraser University, 8888 University Drive, Burnaby, BC V5A 1S6 CANADA 2 Department of Psychology University of British Columbia, 2136 West Mall, Vancouver, BC V6T 1Z4 CANADA Abstract The present study investigated attentional optimization in participants learning rule-based (RB) and information integration (II) categories. Using an eye-tracker to measure the deployment of overt attention, we tracked participants’ learning and optimization during a category learning experiment. We also measured working memory span. We found that participants in the RB condition optimized attention less than II participants before reaching the learning criterion, but more than II participants after criterion, and confirmed that this effect was not due to differences in speed of learning or accuracy. Working memory span was negatively related to pre-criterion optimization in both conditions, but was unrelated to post-criterion optimization. These results show that attentional optimization is influenced by the kind of task being learned or the types of strategies that these tasks elicit, and provide evidence that executive attentional factors influence overt attentional optimization. Keywords: attention, category learning, categorization, rule- based, information-integration, eye-tracking, working memory. Introduction The ability to preferentially process relevant information is critical to achieving effective and efficient performance on virtually any task. Theories of category learning have long recognized the importance of incorporating selective attention into their frameworks, usually simulating selective attention with weights that modulate the importance of stimulus dimensions (e.g. Kruschke, 1992). However, the goal of modeling selective attention has been complicated by the fact that attention is difficult to measure. Some studies have attempted to measure attention by using specially-chosen transfer stimuli to gauge the importance of each stimulus dimension on the categorization decision (e.g., Blair & Homa, 2005). One disadvantage of an indirect measure like this is that it may lead to improper inferences about attentional allocation – for instance, transfer and training may be treated differently by participants (Blair & Homa, 2003). Attentional allocation has also been investigated using a paradigm in which participants use a mouse click to reveal information that they wish to view (e.g., Matsuka & Corter, 2008). This method illuminates exactly which dimensions participants judge to be important, and the order in which they are accessed, but because information that is revealed remains available, no real-time information about which stimulus dimensions participants are considering is recorded. A promising alternative is eye-tracking; it provides fine- grained temporal and spatial information, and is a precise and direct measure of one important aspect of attention: overt attention. Further, there is important overlap between the attentional biases suggested in computational models of attention and participants’ real-life deployment of gaze (Rehder & Hoffman, 2005a). Rehder and Hoffman (2005b) have demonstrated a correspondence between the amount of time participants fixate stimulus features and the value of attention weights generated by model fits of the response data. Kruschke, Kappenman, and Hetrick (2005) showed that measures of eye-gaze were meaningful indicators of attentional flexibility and matched modeling analyses even at the level of individual participants. Eye-tracking studies are beginning to elucidate the role of overt attention in categorization. For example, many important models of categorization assume that attentional weights are task-specific. Blair, Watson, Walshe, and Maj (2009) provided eye-tracking evidence that overt attention can be deployed differentially for different stimuli, supporting a more flexible implementation of attention, like those in recent models (e.g., Kruschke, 2001). In another example, Watson and Blair (2008) used eye-tracking to study participants’ processing of feedback. They found that participants who successfully learned a categorization task spent far more time looking at the re-presented stimulus on incorrect trials than on correct trials, whereas non-learners showed no difference. In contrast, most current theories posit that the re-presented stimulus plays no role in the learning process. Recent progress has also been made in understanding how the allocation of selective attention is optimized during rule-

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.