From Observed Action Identity to Social Affordances

Abstract

Others' observed actions cause continuously changing retinal images, making it challenging to build neural representations of action identity. The monkey anterior intraparietal area (AIP) and its putative human homologue (phAIP) host neurons selective for observed manipulative actions (OMAs). The neuronal activity of both AIP and phAIP allows a stable readout of OMA identity across visual formats, but human neurons exhibit greater invariance and generalize from observed actions to action verbs. These properties stem from the convergence in AIP of superior temporal signals concerning: (i) observed body movements; and (ii) the changes in the body-object relationship. We propose that evolutionarily preserved mechanisms underlie the specification of observed-actions identity and the selection of motor responses afforded by them, thereby promoting social behavior. Combining Observed Body Movements and Objects Changes: The Action's IdentityManual skills are a hallmark of primates, particularly humans. They have made possible most of our transformational impact on the world, which was driven by an evolutionarily preserved but expanding network of cortical areas in the primate lineage that subserves the neural control of manipulative actions [1][2][3][4]. Interestingly, an equally well-articulated neural machinery is required to resolve the visual complexity of observed manipulative actions (OMAs) (see Glossary) performed by other individuals, because this ability is of critical importance for action planning during social interaction and interindividual coordination [5][6][7]. Indeed, as compared with other complex static visual stimuli, such as objects [8], faces [9,10], others' gaze direction [11], and body posture [12], observed actions of others are inherently dynamic stimuli, and their dynamics are essential for an observer's brain to compute their identity, despite the rapid changes in their retinal image. This is probably the reason why James Gibson claimed that 'animals are by far the most complex objects of perception that the environment presents to an observer' [13].

A computational neuroethology perspective on body and expression perception

“…Midlevel features are also different from low-level features (e. g., spatial frequency, luminance). This functional definition of body representation builds on findings in nonhuman species showing that many behaviors cluster in some basic functional domains [15,62]. Currently, a major obstacle for understanding midlevel feature processing is the lack of an analytical and quantitative model of body posture and movement perception.…”

Local and system mechanisms for action execution and observation in parietal and premotor cortices

“… 4 , 12 More recently, two other AON areas have attracted increasing interest: the anterior intraparietal area (AIP) and the pre-supplementary area F6. AIP plays a role in routing to F5 visual information regarding manipulative actions of other 13 , 14 , 15 and area F6 hosts neurons that selectively encode actions and targets of self and others. 11 , 16 , 17 , 18 , 19 Despite these recent advances in our understanding of the AON, two critical questions remain unanswered.…”

The unique role of parietal cortex in action observation: Functional organization for communicative and manipulative actions

“…There is a growing body of research in action observation that supports the hypothesis of an anatomical segregation in parietal cortex for different action classes , including manipulation, interpersonal interaction, skin displacement, locomotion, and climbing ( Jastorff et al., 2010 ; Abdollahi et al., 2013 ; Ferri et al., 2015 ). Action class refers to a group of exemplars that share similar sensorimotor transformations and motor goals ( Orban et al., 2021 ). For instance, grasping or pushing an object are exemplars of the manipulation class, including actions aiming to move or deform an object.…”

“…Similarly, walking and running are exemplars of the locomotion class, including actions aiming to move the body of the actor in space. Note that for observed actions of others, both action class and exemplar are defined as the category levels of observed action identity , which refers to the observable features of an action that distinguish it from other actions ( Orban et al., 2021 ). Observed action exemplars can be characterized as the ordinate level of the observed action identity whereas observed action classes can be characterized as the superordinate level.…”

“…The present study also aims to shed light on this issue with a specific focus on how observed action classes, hence action identity, are represented at the three levels of the AON. By definition, action identity involves two components: 1) body movements of the actor, 2) the changes they produce in the environment, whether it is an object, conspecific, or a substrate ( Orban et al., 2021 ), i.e. the goal of the action.…”