FORCE GRADIENTS IN STIMULUS GENERALIZATION.

COLWINOGRAD (1965) looked at how hard pigeons peck when the color of the key changes. The birds first learned to peck one color for food. Then the team tested many colors the bird had never seen.

They recorded two things: how fast the bird pecked and how hard each peck was. The test colors ranged from the training color to very different shades.

Peck force dropped off in a smooth curve as the color moved away from the training shade. The steepness of that curve depended on how hard the bird had to peck in training. Birds that ended training with soft pecks had steep drop-offs. Birds that ended with hard pecks had flatter curves.

Peck rate also dropped off, but its curve stayed the same steepness no matter the final training force.

Skinner et al. (1958) had already shown that wheel-running rate can be a clean operant unit. COLWINOGRAD (1965) adds force to the list of measurable dimensions, proving it generalizes like rate does.

Storm (2000) later found that response-rate numbers shift when schedule order changes. That work extends L’s basic finding by showing these measures are not fixed; they move with context.

Davison (1992) modeled how variable-interval schedules feed back into behavior. L’s data give real curves that any future model must fit, linking abstract equations to actual force and rate drops.

If you track only rate, you miss half the story. Response force also spreads across new stimuli, and its slope tells you how intense the original requirement was. When you shape a gentle touch or a firm press, expect the force gradient to protect that topography in new settings. Use this to check if your training criteria are too high or too low before you move to novel materials.