Bottom-up processing of curvilinear visual features is sufficient for animate/inanimate object categorization

Abstract

Animate and inanimate objects differ in their intermediate visual features. For instance, animate objects tend to be more curvilinear compared to inanimate objects (e.g., Levin, Takarae, Miner, & Keil, 2001). Recently, it has been demonstrated that these differences in the intermediate visual features of animate and inanimate objects are sufficient for categorization: Human participants viewing synthesized images of animate and inanimate objects that differ largely in the amount of these visual features classify objects as animate/inanimate significantly above chance (Long, Stormer, & Alvarez, 2017). A remaining question, however, is whether the observed categorization is a consequence of top-down cognitive strategies (e.g., rectangular shapes are less likely to be animals) or a consequence of bottom-up processing of their intermediate visual features, per se, in the absence of top-down cognitive strategies. To address this issue, we repeated the classification experiment of Long et al. (2017) but, unlike Long et al. (2017), matched the synthesized images, on average, in the amount of image-based and perceived curvilinear and rectilinear information. Additionally, in our synthesized images, global shape information was not preserved, and the images appeared as texture patterns. These changes prevented participants from using top-down cognitive strategies to perform the task. During the experiment, participants were presented with these synthesized, texture-like animate and inanimate images and, on each trial, were required to classify them as either animate or inanimate with no feedback given. Participants were told that these synthesized images depicted abstract art patterns. We found that participants still classified the synthesized stimuli significantly above chance even though they were unaware of their classification performance. For both object categories, participants depended more on the curvilinear and less on the rectilinear, image-based information present in the stimuli for classification. Surprisingly, the stimuli most consistently classified as animate were the most dangerous animals in our sample of images. We conclude that bottom-up processing of intermediate features present in the visual input is sufficient for animate/inanimate object categorization and that these features may convey information associated with the affective content of the visual stimuli.

Trial registration: ClinicalTrials.gov NCT00001360.

Figures

Figure 1

Three example inanimate objects used in Long et al. (2017) together with (a) their corresponding texform images, created using the Freeman and Simoncelli (2011) algorithm, (b) their corresponding synthesized images created using the Portilla and Simoncelli (2000) algorithm. The images under the “original,” “controlled,” and “texform” columns were directly extracted from figure 1 of Long et al. (2017). The images under the “texform” column were created using the algorithm described in Freeman and Simoncelli (2011) with some slight modifications outlined in Long et al. (2016). The images under the “synthesized” column were created by using the algorithm described in Portilla and Simoncelli (2000), which we used in this study. Both the “texform” and “synthesized” object algorithms used the images under the “controlled” column as inputs.

Figure 2

Example stimuli and sample trial displays from the rating and classification sessions of the experiment. (A) Examples of animate images with their corresponding synthesized stimuli. (B) Examples of inanimate/man-made images with their corresponding synthesized stimuli. (C) A sample trial display from the rating session of the experiment. This example depicts a trial in which a participant rated the synthesized stimuli on degree of “boxiness” or how rectilinear the images appeared to him or her. The black bar corresponds to the amount of “boxiness” this participant attributed to the stimulus image. (D) Sample trial display from the classification session of the experiment. This example depicts a trial in which a participant classified the synthesized stimulus as “animate.” The black bar represents the confidence of the participant on his or her classification choice.

Figure 3

The figure depicts how the classification accuracy of the animate (X) and inanimate (O) images varied as a function of the amount of calculated curvilinear information present in the images. The best-fit line for the animate category is represented by the solid black line, and the best-fit line for the inanimate category is represented by the dotted gray line. The average classification accuracy was 54.62%, which is significantly above chance.

Figure 4

Rank-ordered classification accuracy for the synthesized animate images in descending order. A sample of the original images, corresponding to the rank-ordered image IDs of the synthesized images, are presented on top of the bars. The horizontal dashed line represents chance performance (50% accuracy). The error bars denote ±1 SEM.