Neuromorphic models of response time

A new approach to modelling response time is proposed, based on the activation dynamics of simplified neural units. The proposed neuromorphic decision units are dissipative and have bounded activation, as do real neurons. A decision is made either when the unit's activation exceeds a threshold or when it converges, using a criterion based on the derivative of activation. First, the relationship between varieties of neuromorphic and sequential sampling models is reviewed. Mathematical and simulation results are presented both for deterministic and random versions of the neuromorphic units. These results are used to highlight strengths and weaknesses of this new approach to dynamic models of decision making.