The key concepts that have driven my curiosity in research are: neural networks, self-organization, reinforcement learning, language acquisition, meaning representation, grounded cognition, cognitive robotics, causality, and others.
Some time ago, I did work in the field of early lexical acquisition using self-organized networks. The focus was on unsupervised approach that accounts for several developmental phenomena observed in young children.
The issue of (syntactic and semantic) systematicity is a continuing challenge for connectionist models that have the ambition to be a plausible account of human cognition. In our work, we focused on syntactic systematicity, tested in a new neural network model (RecSOMsard), as well as the Recursive Auto-Associative Memory (RAAM).
Recurrent self-organizing maps
We theoretically and experimentally investigated the RecSOM model, and experimentally compared it with other two models (RSOM and SOMSD) on three data sets with different complexity. These models can be used for representing sequential data, yielding the topographically organized receptive fields of neurons.
Causality in mind-body relationship
I have for long been fascinated by the "old" problem of the relation between the material body and immaterial mind. On of the tricky questions in this context is the problem of mental causation. I wrote some papers about it that reflect my view about the role of the mental in the physical world, given the physical closure assumption.
Cognitive developmental robotics
We use the simulated iCub robot in the effort to model the acquisition of various early cognitive skills, such as visuomotor coordination and object-directed actions.
Neural correlates of action
We started to design EEG experiments focusing on measuring neural correlates of observed motor action and motor imagery. This research is closely related to the mirror neuron system that is assumed to underlie the human ability to understand observed motor behavior by linking perception and action.