Reinforcement learning - help building a model

Thanks for the reply.
I was actually wondering whether I need to test logp (like your code) instead of estimating the SCR response directly. Although, I’m having a bit of trouble revising your code to multiple subjects (in theano).