Now you see it, now you don't: The context dependent nature of category-effects in visual object recognition

Abstract

In two experiments, we test predictions regarding processing advantages/disadvantages for natural objects and artefacts in visual object recognition. Varying three important parameters—degree of perceptual differentiation, stimulus format, and stimulus exposure duration—we show how different category-effects can be provoked in normal subjects on the same task. We interpret the results in light of the Pre-semantic Account of Category Effects (PACE; Gerlach, 2009), and conclude that category-effects do not reflect absolute processing differences between categories. Rather, category-effects are products of common operations which are differentially affected by the structural similarity among objects (with natural objects being more structurally similar than artefacts). The potentially most important aspect of the present study is the demonstration that category-effects are very context dependent; an aspect which has been neglected in earlier literature on category-specificity. PACE provides a useful framework in this respect, as it specifies how category-effects are influenced by changes in task and stimulus characteristics.

Keywords:

• Category-specificity
• Global shape
• Perceptual differentiation
• Shape configuration
• Visual agnosia

Acknowledgements

We would like to thank Troels Tarp for collecting the data for Experiment 1, F. F. Fakutsi for invaluable comments and inspiration, and Randi Starrfelt for useful comments on an earlier version of this manuscript.

Notes

¹Due to the repetition of items within subjects we initially tested whether there was evidence of presentation order effects on RTs and error rates. This was done by collapsing the data for all four subjects having performed the same task at the same point in time; that is, as either the first, the second or the nth task. Hence, for each task we had six data points each representing the average of the four subjects who performed the task as the nth task. We then computed the correlation between presentation order and RT/error rate for each of the six tasks individually. No significant effects of presentation order were revealed in the RTs. However, we did find effects of presentation order on error rates, with the error rate decreasing as a function of repetition, for the following three tasks: Easy object decision with silhouettes (Pearson's r=−.92, p=.01); Difficult object decision with silhouettes (Pearson's r=−.93, p<.01); and Difficult object decision with fragmented forms (Pearson's r=−.86, p<.05). Despite the fact that repetition generally seems to reduce the error rate for degraded stimuli (silhouettes and fragmented forms) but not for full line drawings, the error rate is still significantly higher for degraded drawings compared with full line drawings (as will be described in the main text). Accordingly, while repetition may have reduced differences between degraded drawings and full line drawings, the degradations performed are still effective.

²Note that the Snodgrass and Vanderwart (1980) measure of visual complexity is based on each stimulus as a whole (both global shape and internal details) and that our selection of artefacts and natural objects did not differ significantly in this measure.

³We also tested whether there were effects of presentation order on RTs/error rates in Experiment 2, using the same procedure as described for Experiment 1. No significant effects of presentation order were revealed in the RTs. However, we did find effects of presentation order on error rates, with the error rate decreasing as a function of repetition, for the following three tasks: Difficult object decision with outline drawings (Pearson's r=−.96, p<.01); Easy object decision with silhouettes (Pearson's r=−.95, p<.01); and Difficult object decision with fragmented forms (Pearson's r=−.86, p<.05). Despite this, and as described in the main text, we still find: (i) that more errors are made with silhouettes than with full line drawings in the easy tasks; (ii) that more errors are made in the difficult compared with the easy tasks with full line drawings; and (iii) that more errors are made in the difficult compared with the easy tasks with fragmented forms. Accordingly, while repetition may have reduced differences between easy and difficult versions of object decision with full line drawings and fragmented forms, and between easy object decisions with silhouettes compared with full line drawings, effects of difficulty and stimulus type are still present.

⁴Our expectation regarding the degree of perceptual differentiation required for solving the easy object decision task with full line drawings was based on the assumption that easy object decisions require less perceptual differentiation than object naming at the basic-level. As category-effects in object naming paradigms are often rather small and, as mentioned in the introduction, inconsistent, we therefore expected no category-effects during easy object decision tasks. Prompted by the results of Experiment 1, we have analyzed unpublished functional imaging data from our laboratory, looking at activations in occipitotemporal areas of the brain during easy object decisions and basic level naming. This analysis revealed that occipitotemporal activations are significantly greater during easy object decisions than during object naming at the basic level, suggesting that the degree of perceptual differentiation needed for solving the easy object decision task may indeed be higher than we first assumed.