Emergence of Mirror Neurons in a Model of Gaze Following

Abstract

Gaze following is the ability to redirect one's gaze to the location where another agent is looking. We present a computational model of how human infants or other agents may acquire gaze following by learning to predict the locations of interesting sights from the looking behavior of other agents through reinforcement learning. The model accounts for many findings about the development of gaze following in human infants. During learning, the model develops pre-motor representations that exhibit many properties characteristic of mirror neurons, but they are specific to looking behaviors. The existence of such a new class of mirror neurons is the main prediction of our model. The model also offers a parsimonious account of how these and possibly other mirror neurons may acquire their special response properties. In this account, visual representations of other agents' actions become associated with premotor neurons that represent the intention to perform corresponding actions. The model also demonstrates how this development may be obstructed in autism spectrum disorder, giving rise to specific physiological and anatomical differences in the mirror system.

A unified account of gaze following

“…Our model previously predicted the emergence of a new class of mirror neurons for gaze-following emerging in its premotor layer [66]. 3 Such kinds of mirror neurons have indeed been recently found in macaque monkeys 4 [58].…”

“…This has allowed us to replicate a wealth of gaze following phenomena, as discussed throughout the paper. 2 Earlier versions of this model were presented in [37], [39], and [66], with newer versions successfully replicating results for more experimental setups.…”

Higher order thoughts in action: consciousness as an unconscious re-description process

“…Social context as a driving force for learning has, indeed, been recognized in language learning [59], child development [60] and social cognition [61]. Recently, it has also been suggested that mirror neurons might be the result of reinforcement learning [62][63][64]. Thus, the third loop conceptualizes metacognition as resulting from predictive learning mechanisms that allow for agents to simultaneously learn about the environment as well as about their own internal representations.…”

Joint Attention by Gaze Interpolation and Saliency

“…Most of the joint attention approaches in the literature do not explicitly correct for the discrepancy between the head pose and gaze direction, which is reported to be normally distributed with a mean of five degrees in natural settings [36], [37]. In computer graphics literature, there are some recent approaches that model this discrepancy explicitly, in order to synthesize virtual agents with a more natural behavior [30], [38].…”