Human performance on negative slope schedules of points exchangeable for money: a failure of molar maximization.
Humans on negative-slope point schedules show burst-pause patterns, not the steady low rate maximization theories predict.
01Research in Context
What this study did
Nasr et al. (2000) asked adults to press a button for points that turned into cash. The schedule had a negative slope: more presses lowered the cash value. They tracked how fast people pressed and compared it to what strict maximization or matching theories predict.
Each person worked alone in a quiet room. The computer tallied points and showed the exchange rate in real time. The team ran the schedule for many sessions to see stable patterns.
What they found
People did not maximize money. They pressed in quick bursts, then paused. Their overall rate stayed higher than the theories said it should be.
The burst-pause pattern showed up in every participant. It did not smooth out with practice.
How this fits with other research
Hopkins et al. (1977) saw the same kind of undermatching in pigeons on concurrent VI-VI keys. Both studies show that strict matching math misses real behavior.
Calamari et al. (1987) already weakened molar maximization in pigeons choosing between FR and PR schedules. Nasr et al. (2000) now show the idea also fails with humans on negative slopes.
Webb et al. (1999) proved that deprivation shifts the asymptote of Herrnstein’s hyperbola, another crack in matching theory. Nasr et al. (2000) add a new crack: schedule direction matters, not just rate.
Why it matters
When you write token boards or point systems, do not assume clients will automatically settle into the most efficient rate. Expect bursts and pauses, especially if faster responses lower reward value. Use visual feedback and paced instructions to shape smoother, more cost-effective responding.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Add brief pause prompts to your token app to slow burst responding and protect reward value.
02At a glance
03Original abstract
Panel pressing was generated and maintained in 5 adult humans by schedules of points exchangeable for money. Following exposure to a variable-interval 30-s schedule and to a linear variable-interval 30-s schedule (which permitted points to accumulate in an unseen "store" in the absence of responding), subjects were exposed to a series of conditions with a point-subtraction contingency arranged conjointly with the linear variable-interval schedule. Specifically, points were added to the store according to the linear-variable interval 30-s schedule and were subtracted from the store according to a ratio schedule. Ratio value varied across conditions and was determined individually for each subject such that the subtraction contingency would result in an approximately 50% reduction in the rate of point delivery. Conditions that included the subtraction contingency were termed negative slope schedules because the feedback functions were negatively sloped across all response rates greater than the inverse of the variable-interval schedule, in this case, two per minute. Overall response rates varied inversely with the subtraction ratio, indicating sensitivity to the negative slope conditions, but were in excess of that required by accounts based on strict maximization of overall reinforcement rate. Performance was also not well described by a matching-based account. Detailed analyses of response patterning revealed a consistent two-state pattern in which bursts of high-rate responding alternated with periods of prolonged pausing, perhaps reflecting the joint influence of local and overall reinforcement rates.
Journal of the experimental analysis of behavior, 2000 · doi:10.1901/jeab.2000.73-241