Visual imagery of faces and cars in face-selective visual areas

Abstract

Neuroimaging provides a unique tool to investigate otherwise difficult-to-access mental processes like visual imagery. Prior studies support the idea that visual imagery is a top-down reinstatement of visual perception, and it is likely that this extends to object processing. Here we use functional MRI and multi-voxel pattern analysis to ask if mental imagery of cars engages the fusiform face area, similar to what is found during perception. We test only individuals who we assumed could imagine individual car models based on their above-average perceptual abilities with cars. Our results provide evidence that cars are represented differently from common objects in face-selective visual areas, at least in those with above-average car recognition ability. Moreover, pattern classifiers trained on data acquired during imagery can decode the neural response pattern acquired during perception, suggesting that the tested object categories are represented similarly during perception and visual imagery. The results suggest that, even at high-levels of visual processing, visual imagery mirrors perception to some extent, and that face-selective areas may in part support non-face object imagery.

Fig 1

Schematic of scanning procedure for a sample (a) perception run and (b) imagery run. In perception runs, participants indicated via key press whether the image was displaced upward or downward.

Fig 2. Example 27 functional voxel ROIs shown on 2 inflated right hemispheres of representative participants (anterior and posterior portions of the brain denoted with A and P, medial and lateral with M and L).

Note: All ROIs were defined in volumetric space. ROIs transformed to surface space for illustrative purposes only. Any dis-contiguous clusters or ROI overlap that appears in figure did not occur in volumetric space (all ROIs were contiguous and no ROIs contained overlapping voxels).

Fig 3

Average accuracies of the classifier when trained on imagery data and tested on imagery data for object versus car (left) and face versus object (middle) and face versus car (right) two-way classifications in face-selective ROIs (upper row) and object-selective ROIs (lower row). Error-bars represent one-tail 95% confidence interval, accuracy of decoding below chance was not theoretically meaningful.

Fig 4

Average accuracies of the classifier when trained on imagery data and tested on perception data for object versus car (left) and face versus object (middle) and face versus car (right) two-way classifications in face-selective ROIs (upper row) and object-selective ROIs (lower row). Error-bars represent one-tail 95% confidence interval, accuracy of decoding below chance was not theoretically meaningful.

Fig 5

Average parameter weights for faces (dashed), objects (dotted) and cars (solid) plotted for perception runs 1 and 2 in combined face-selective regions (right, significant run*category interaction) and object-selective regions (left). Fixation is used as a baseline for the face, object and car conditions and error bars show SEM values.