Generalization of conditioned suppression after differential training.

The team trained rats to tell two click speeds apart. One speed meant mild shock was coming. The other speed meant no shock.

After many trials they tested new click speeds in between. They tracked how much each rat froze to see where fear spread.

Fear did not spread evenly. Most rats peaked on the side of the safe cue, not the danger cue. Each rat had its own lopsided curve.

The shape of the curve changed from rat to rat, but the shift away from danger was steady.

WINOGRAD (1965) used the same clicks and shocks but tested many speeds at once. That work showed smooth bowl-shaped curves. P et al. added differential training and revealed the sideways shift.

Neuringer (1973) saw flat curves after simple presence-absence training in pigeons. The difference is training style: one cue versus two cues along the same sound dimension. Rats with clear S+ and S- show the tilt; pigeons without that contrast do not.

Grosch et al. (1981) later got the same sideways peak under avoidance. Together the studies say: when you reinforce two spots on a stimulus line, responding slides away from the bad spot, no matter if the reinforcer is food, shock, or escape.

If you teach a client to tell two tones apart, expect generalization to skew away from the punished tone. Probe more values on the safe side before you call mastery. Also, mix in many S- examples early; the literature shows that single-cue training flattens control. Use the asymmetry to your advantage when shaping new discriminations.