Theoretical interpretations of a Markov model for avoidance conditioning