From single cells to social perception

Abstract

Research describing the cellular coding of faces in non-human primates often provides the underlying physiological framework for our understanding of face processing in humans. Models of face perception, explanations of perceptual after-effects from viewing particular types of faces, and interpretation of human neuroimaging data rely on monkey neurophysiological data and the assumption that neurophysiological responses of humans are comparable to those recorded in the non-human primate. Here, we review studies that describe cells that preferentially respond to faces, and assess the link between the physiological characteristics of single cells and social perception. Principally, we describe cells recorded from the non-human primate, although a limited number of cells have been recorded in humans, and are included in order to appraise the validity of non-human physiological data for our understanding of human face and social perception.

Figure 1.

Illustration of some cortical regions in the monkey shown to contain face-sensitive cells (adapted from Seltzer & Pandya [25]). The STS has been opened out in order to illustrate regions within the sulcus. Inferotemporal cortex (IT) consists of regions on the surface of the temporal lobe TE3 and TEm that also extends into the lower bank of the STS. TEa is entirely situated on the lower bank. The floor of the superior temporal sulcus (FST) is usually considered part of the motion-processing system (including also MT/V5 and the medial superior temporal sulcus, MST), and to our knowledge contains no face-sensitive cells. Still within the floor of the STS, but more anterior to FST is IPa. PGa extends anteriorly within the floor of the STS. The upper bank of the STS consists of two elongated regions, TPO and TAa. Orbitofrontal cortex (OFC) and ventro-lateral pre-frontal cortex (vlPFC) also contain face-sensitive cells, as well as the anygdala, located within the pole of the temporal lobe, but not seen in this illustration.

Figure 2.

Responses of one cell selective for faces. Responses to more than a thousand images were compared with a progressive narrowing of the collection of images keeping the most effective 75% and the least effective 25% stimuli [24]. After a series of trials, the images were ranked in effectiveness and the 40 most effective stimuli all contained faces. Given the proportion of test images containing faces, the probability of obtaining this pro-face selectivity by chance would be one in 1 000 000 000 000.

Figure 3.

Latency ranges for face-sensitive cells recorded in different cortical areas in the monkey. Asterisk denotes that the estimate includes cells recorded with selectivity for other non-face objects.

Figure 4.

Images of different facial expressions performed by one individual macaque. (a) Coo—an affiliative call often used to contact other individuals over short and medium distances, characterized by a pursing and protuberance of the lips. (b) Grunt—a soft sounding call associated with positive interactions and food, often accompanied by blinking. (c) Licking lips—in the absence of food, associated with grooming. (d) Lip-smacking—a repetitive subordinate gesture used to appease more dominant individuals. (e) Pant threat—a call used as a threat, characterized by opening of the mouth showing teeth and staring. (f) Yawn—opening of the mouth and showing of the teeth, raising of the head and often eyelid closure.

Figure 5.

An example, superior temporal sulcus neuron that codes the audiovisual oro-facial communicative gesture ‘coo’. Rhesus macaques use coos to communicate over varying distances with allies. There is a small visual response (red), and very little auditory response (orange). When the visual and auditory signals are presented simultaneously, there is a marked augmentation of the response (green). Pant–threat vocalizations (a threatening signal to a macaque) result in a poor response (blue), and when paired with the sight of the coo oro-facial gesture do not result in much of an increase in the cell response (purple). Responses are spike density waveforms with 12 ms Gaussian smoothing; the shaded areas indicate the standard error of the mean. The data are adapted from Barraclough et al. [119]. Red line, coo vision; orange line, coo sound; green line, coo vision + coo sound; blue line, pant–threat sound; purple line, coo vision + pant–threat sound.