Social Groups as a Mental Category

Mersad and Caristan [1] used nested complex stimuli: a dyad composed of two close but separated busts, each of them being itself a complex stimulus formed with parts (head and shoulders). In this case, the evidenced perceptual integration reflects indubitably high-level visual computations. However, what is exactly the nature of the integration process? A parallel with Boremanse et al. [2] study where the authors reported neural integration in a closely similar paradigm but with different stimuli, might afford insight into this question. In their study, these authors could objectively dissociate the neural response of each half of a face, reflected in the input frequencies, from the response to the face as a whole, which appeared as a combination of these input frequencies. This latter response provided a signature of non-linear integration of face halves into a perceptual unit, comparable to the neural convergence observed for the two silhouettes stimuli in Mersad and Caristan [1]. However, the nature of the stimuli constituting the parts is a critical difference between the two studies. Indeed, the half of a face do not belong to the realm of natural visual entities: we often perceive faces in profile, but we rarely see only the half of a face. Thus, perceptual integration of face halves into a coherent unit, observed in Boremanse et al. [2] study, is a form of (perceptual) reconstruction of an object, unnaturally divided in two parts presented at different temporal frequencies to be dissociated.

Moreover, the face halves were contiguous which could have induced low-level neural integration as the small receptive fields spanning the border of the two halves could be responsible for generating responses at IM components.

Unlike the half of a face, a human silhouette is an independent visual entity that is part of the ecological world. In Mersad and Caristan [1], two silhouettes were presented close to each other in space but with a clear gap between them. Here, response at IM frequency signals the existence of a neural network processing both of these physically distant visual items, joining them into a coherent percept [3].

Critically, the IM response was significantly reduced when human silhouette stimuli were replaced with another category of objects, two silhouettes of chairs, and when the dyad was presented upside down, an inversion effect observed in visual perceptual categorization of objects [4].

Our proposal is that the IM component reflecting perceptual integration evidenced in Mersad and Caristan [1] experiment is a neural marker of mental categorization of a visual stimulus as a group of humans (two here) represented as an entity per se.

Perception of Nested Structures: Global or Local Level?

An enduring question in psychology that can be tackled by the Frequency Tagging approach is the relations between whole (global) and part (local) levels of processing, an issue discussed in Mersad and Caristan [1] study. The authors noted that while inversion effect was observable at the global dyad level, it was absent from the local silhouette level, and this, in spite of abundant data in the literature showing that body pictures and silhouettes are subjected to inversion effect [5].

The dominance of global visual percept over local information has been documented from early Gestalt theories [6] and Navon’s ‘Forest before trees’ experiment, to ensemble coding studies showing that a summary representation is favored over the representation of local details [7]. These studies showed that visual percepts are dominated by global information using paradigms such as shape detection and identification [8], processing time of gist extraction from visual scenes [9] or statistical summary ensemble computations [7]. Some studies highlighted the role of structural or configural information, for example, Haberman et al., [7] found a decline in the ability to compute the average emotional expression of a set of faces when the faces where presented upside down.

However, Mersad and Caristan [1] could experimentally dissociate the neural entrained response to the whole-dyad silhouette from the response to the individual silhouettes forming the dyad, while presenting these stimuli either in (typical) upright or (altered) inverted position (Figure 1). Thus, the authors could compare the processing of the structural information between the two levels (local and global) of the visual perceptual hierarchy. They showed that configural information at the global level dominates perception, and this, specifically when the nested stimuli represented human silhouettes. Hence, their results offer a new kind of demonstration of global primacy.

Perception for Action: When Facing Human Pluralities

Gibson’s ecological theory holds that perception is a guide for adaptive behavior [10]. Indeed, following Gibson’s Perceiving is for doing principle, perception, while offering affordances, is closely linked to action.

Do distinctive mechanisms for perception of human groups constrain individual’s attitudes and actions?

To our knowledge, no answer has been proposed to this question in the literature. However, we can distinguish two different types of processes that has been reported for visual perception of human aggregates: statistical summary principles described in ensemble coding (e.g. Habermann and Whitney, 2012 [7]) and holistic processing highlighted in Mersad and Caristan [1], although in the latter case the plurality was consisting only of two humans.

A tentative framework encompassing these two mechanisms is that distinctive perception of human groups is useful for having access to a picture of the majority trend by generating a compressed representation of human ensembles, while disregarding precise information about individuals. Note that the two mechanisms cited above are likely to be complementary: while summary statistics allows gathering average information with better speed, holistic processing captures properties at the global level that are not reducible to individual levels.

Interestingly, the latter Gestalt perceptual mechanism that captures emergent properties resulting from the interaction of the parts is not without pointing out multi-agent models describing unpredictable emergent behaviors that appears when collectives of agents are in interaction [11].

Our proposal is that holistic perceptual mechanism of human groups might intervenes in the identification and analysis of emergent behavior characterizing human pluralities. Indeed, these are critical abilities for a human being to be able to read intentions at the group level and to tune behavior and action consequently.

Future investigations should shed light on the nature of the link between group perceptual mechanisms and the adaptive action they trigger in the individual, extending the framework to variable sizes of human groups.