The role of category learning in the acquisition and retention of perceptual expertise: A behavioral and neurophysiological study

Abstract

This study examined the neural mechanisms underlying perceptual categorization and expertise. Participants were either exposed to or learned to classify three categories of cars (sedans, SUVs, antiques) at either the basic or subordinate level. Event-Related Potentials (ERPs) as well as accuracy and reaction time were recorded before, immediately after, and 1-week after training. Behavioral results showed that only subordinate-level training led to better discrimination of trained cars, and this ability was retained a week after training. ERPs showed an equivalent increase in the N170 across all three training conditions whereas the N250 was only enhanced in response to subordinate-level training. The behavioral and electrophysiological results distinguish category learning at the subordinate level from category learning occurring at the basic level or from simple exposure. Together with data from previous investigations, the current results suggest that subordinate-level training, but not basic-level or exposure training, leads to expert-like improvements in categorization accuracy. These improvements are mirrored by changes in the N250 rather than the N170 component, and these effects persist at least a week after training, so are conceivably related to long-term learning processes supporting perceptual expertise.

Introduction

Recent studies of perceptual expertise and categorization have used training studies to further our understanding of the behavioral and neural mechanisms contributing to the acquisition of visual perceptual expertise (Gauthier and Tarr, 1997, Gauthier et al., 1999, Gauthier et al., 1998, Rossion et al., 2002, Rossion et al., 2004, Scott et al., 2006, Tanaka et al., 2005). The use of training studies allows for more precise control over the amount and quality of visual experience needed to obtain perceptual expertise. Although researchers do not expect to be able to equate the acquisition of expertise in the laboratory to real-world expertise, training in a laboratory setting allows for better manipulation of factors contributing to perceptual learning and generalization. Results of perceptual training studies have lead to several important conclusions about how people learn to categorize at different levels, how category learning generalizes to novel exemplars and categories, and how this type of learning may be implemented at the neural level (Gauthier and Tarr, 1997, Gauthier et al., 1998, Rossion et al., 2002, Rossion et al., 2004, Scott et al., 2006, Tanaka et al., 2005).

Tanaka, Curran, and Sheinberg (2005) trained participants to classify species of wading birds and species of owls at either the subordinate (species, e.g., Barn owl) or basic (wading bird) level of abstraction. Training took place on 6 days over a 2-week period with the amount of training trials equated for both subordinate and basic-level conditions. Behavioral results of this study suggest that subordinate, but not basic-level training, increased discrimination of previously trained birds. Moreover, greater generalization to novel exemplars within trained species, and novel exemplars of untrained species (within the same family) was found for subordinate compared to basic-level training. These data suggest subordinate-level discrimination training is an important factor in the acquisition of perceptual expertise and the subsequent transfer to new exemplars from learned categories and new exemplars belonging to novel, but structurally related categories.

In a recent follow-up study, Event-Related Potentials (ERPs) were recorded before and after training at the subordinate and basic levels (Scott et al., 2006). The behavioral results of this investigation replicated previous findings suggesting subordinate but not basic-level training led to increased discrimination of trained birds and increased generalization of untrained birds. We also identified two distinct ERP components, the N170 and the N250, that were correlated with the acquisition of perceptual expertise. Whereas the N170 was sensitive to the encoding of basic-level, shape information, the N250 was modulated by the more fine grain perceptual detail required for subordinate-level identification (Scott et al., 2006, Tanaka et al., 2006). Generalization to untrained exemplars was also found for both the N170 and the N250 components. These results suggest that increased discrimination and generalization required for subordinate-level judgments map more directly onto the N250 component than the N170 component, which has been previously associated with real-world expertise (Tanaka and Curran, 2001, Gauthier et al., 2003). In addition, these data further question the notion that the N170 specifically indexes face processing (Carmel and Bentin, 2002, Eimer, 2000, Sagiv and Bentin, 2001) and instead provide additional evidence for a more general experience based N170.

The present investigation sought to further clarify the factors contributing to the acquisition of perceptual expertise, including the function of the ERP components correlated with categorization and perceptual expertise. This research addresses three unanswered questions. First, how does learning, mediated by tasks including feedback and category labeling, differ from exposure-only learning? Previously, we found both behavioral and electrophysiological differences for categories of birds trained at the subordinate versus the basic level (Scott et al., 2006). Here we extend this finding and examine whether subordinate- and basic-level learning contribute anything above and beyond simple exposure learning. Second, for how long after training are behavioral and electrophysiological training effects maintained? More specifically, is it necessary for training to continue in order to sustain the increases in performance and ERP amplitude we previously reported (Scott et al., 2006)? If the effects of training are short-lived in this paradigm, we must consider the relevance of these results to real-world perceptual expertise. Finally, does training with cars, an artificial (as opposed to a natural kind) object category, influence learning and generalization across different levels of training? Multiple exemplars of multiple models of three different types of cars were used as experimental stimuli. Car stimuli were used to first determine whether or not we could replicate and extend our previous results with a new class of stimuli. Furthermore, we wanted to establish whether learning objects from a human-made category, such as cars, yielded similar or different results than from learning objects from a natural category, such as birds.