[not in handout, see intranet
or see Lieberman (2000) page 241]

Colwill, R.M. and Rescorla, R.A. (1985) Post- conditioning devaluation of a reinforcer affects instrumental responding. Journal of Experimental Psychology: Animal Behaviour Processes, 11, 120-132.

3 expts with rats, using a Skinner box with a bar on the left and a chain on the right, with food magazine in the middle which could deliver either sucrose solution or standard pellets, the animals being trained using a different reward for each response.

Subsequently, only one kind of reward was paired with lithium chloride injections. (Or, in expt 3 animals were given free access to one kind of reward prior to testing.)

Then animals were tested in extinction (no rewards of either kind).

Results The animals made few responses of which ever kind had produced that reinforcer which had been devalued, but many responses of the kind trained with the other reward

"The results are interpreted as showing that a reasonably detailed representation of the reinforcer is encoded in instrumental learning." (Abstract.)

Colwill and Triola (2002) found that rats extensively trained when hungry with a response (e.g. chain pulling) rewarded with a weak sugar solution, and a different response (lever pressing) rewarded with food pellets showed a strong preference for making the response previously rewarded with the liquid when tested for 10 minutes when thirsty.


[not in handout, see intranet]

Rescorla, R.A. (1991) Associative relations in instrumental learning: The eighteenth Bartlett Memorial Lecture. Quarterly Journal of Experimental Psychology, 43B, 1-23.

  • He reviews a variety of experiments, supporting the existence of binary associations, between any two of stimuli, responses, and response outcomes.

  • The contrast with Thorndike's Law of Effect is that responses can be associated with their consequences.

  • He also supports a “hierarchical structure in the form S®(R®O)” [a response can be associated with its usual outcome in the presence of a prior stimulus — this is exactly what was proposed by Tolman (1959)]