Adaptive changes in response to extinction can be observed if provisions are made that permit behavior to shift away from trained routine. In the present case, the baiting of four arms in the eight-arm radial maze increasingly restricted subject movements to those arms. The unbaited arms afforded a new direction for behavior to take during extinction. Withdrawal of reward was followed by an immediate and active expansion of sites visited. The previously unrewarded arms were now regularly sampled. This was one consequence of extinction. The other was the well-known decline in overall rate of arm entry, whether arms were previously baited or not. Ethanol, at doses of 1.5 or 2.0 g/kg, eliminated the spatial dispersion attendant upon nonreward. It did not affect the decline in overall responsiveness. Ethanol's deletion of extinction-induced spatial variability may account for its impairment on reversal and other kinds of tasks that require a shift away from old patterns of behavior.