Relating computational models and experiments in peripersonal space: Comment on "computational models of peripersonal space representation", by Bertoni et al.

Peripersonal space (PPS) refers to the space immediately surrounding the body, and has been shown over the past half century to be represented by a specialised set of neural mechanisms. Research on PPS has used the full range of methods in neuroscience and psychology, incorporating evidence from neurophysiology [1], experimental psychology [2], neuropsychology [3], neuroimaging [4], and as reviewed by Bertoni and colleagues, computational modelling. As the authors note, computational models have important advantages in forcing researchers to explicitly define the functional properties of the systems they are studying. One main problem with proposing theories without a mathematical formalisation is that they are often ambiguous and hard to compare with other theories. This problem can be avoided with computational models, as formalisation is crucial when modelling. In this way it is easier to identify mathematical equivalencies or errors, and explaining underlying mechanisms [5,6]. This review of computational models of PPS is a timely and welcome addition to the field. This is especially true as computational models are often not easily accessible to researchers working on similar topics using other methods. The authors have made a significant effort to establish organisational principles for the models, which is highly useful for a broad range of researchers, not just those focused on PPS. This approach provides a solid approach to computational modelling and offers a concrete application of Marr ’ s [7] levels-of-analysis framework, which often remains abstract. For researchers working on PPS, it will be highly valuable to have a list of models alongside a framework for interpreting them. Here, we wish to raise some issues relating to the relation between modelling and experimentation, which we believe are important to make the relation between computational models and experimentation on PPS productive and mutually beneficial