The peripersonal space may be considered as the action space within which the body directly interacts with external objects. There is converging evidence that peripersonal space is represented by a specialized circuit of multimodal neurons integrating tactile stimuli applied on a body part with visual stimuli delivered near the same body part, e.g. the hand. Tools used to extend the action space may modify the boundaries of the peri-hand area, where vision and touch are integrated. In extinction patients, a far visual stimulus at the tip of a right-held tool produces left tactile extinction similar to a near visual stimulus at the hand, only after the patients have used the tool to reach the far space. The neural mechanisms underlying such plasticity have not been yet identified. To this aim, neural network modeling may be integrated with the experimental research.
In this work, we pursued two main objectives: i) to use an artificial neural network in order to postulate some physiological mechanisms for peri-hand plasticity able to account for in-vivo data; ii) to validate the artificial network by testing predictions derived from simulations with an ad hoc behavioural experiment on an extinction patient.
The model assumes that the modification of peri-hand space arises from a Hebbian growing of visual synapses converging into the multimodal area, which extends the visual RF of the peripersonal bimodal neurons. Under this hypothesis, the model is able to interpret and explain controversial results in the literature on peri-hand space reconfiguration, ascribing different results to different tool-use tasks during the learning phase. Importantly, model hypothesis implies that, after tool-use, re-coding of a far visual stimulus as a near one should be observed in vivo, even in absence of any physical connection between the subject’s hand and the far space. This prediction has been validated experimentally on a right brain damaged patient with visuo-tactile extinction. Before tool use, a visual stimulus near the right hand induced much more left tactile extinction than a far stimulus; after few minutes of tool use, the amount of visual-tactile extinction at the far location was similar to that at hand location, suggesting an extension of the integrative peri-hand space. Crucially, this effect was analogous both when the patient passively held the tool in her right hand, and when the tool was removed. Computational and experimental results of this study may provide important contribution to gain a deeper insight into the neural and functional mechanisms of peripersonal space representation and its plasticity. This work demonstrates how neural network modelling and simulations may integrate with experimental studies, by generating new predictions and suggesting novel and fruitful experiments to investigate cognitive processes.