Contingency discriminability, matching, and bias in the concurrent-schedule responding of possums (Trichosurus vulpecula).
Reinforcer type nudges choice in possums the same way it does in pigeons and people, and the plain matching law still works if you add a small bias term.
01Research in Context
What this study did
The team worked with possums on two-lever concurrent VI-VI schedules.
They swapped the reinforcer type on one lever while keeping rate the same.
The goal was to see if reinforcer type alone would bias choice and which math model best caught the shift.
What they found
Possums shifted their lever presses when the treat changed.
Both the generalized matching law and a newer model fit the numbers, but the matching law kept its simpler, clearer parameters.
Reinforcer type acts like a built-in bias, similar to what rate or amount does in other species.
How this fits with other research
Macdonald et al. (1973) first showed pigeons match response ratios to reinforcer ratios on the same two-lever setup.
Bron et al. (2003) now says the same rule holds for possums, adding "type bias" to the list of predictable shifts.
Davison et al. (1984) found that mixed reinforcer durations broke the matching law; the possum study answers back by showing type bias does NOT break it—it just slides the line.
Hall (2005) later added that when earning rates differ across keys, matching fails again; together these papers draw the boundary: rate and type tweaks are okay, duration and earning-rate splits are not.
Why it matters
If a client’s choice looks off, check what sits on each side of the equation.
A richer snack, a brighter token, or a louder praise can act like extra reinforcers without changing the schedule.
You can keep the simple matching equation in your toolkit—just add a bias parameter for the reinforcer type and move on.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Spot-check your concurrent reinforcers: if one side offers a preferred item, treat that preference as a built-in bias and adjust the scheduled rate downward to keep response ratios where you want them.
02At a glance
03Original abstract
Six possums (Trichosuruus vulpecula) responded under dependent concurrent variable-interval variable-interval schedules of reinforcement. Over 15 conditions, barley-carob was one reinforcer with the other reinforcer consisting of Coco Pops, coconut, or a barley-carob mixture with 0%, 2%, 4%, or 6% salt added to the barley. The schedules were both variable-interval 40 s. As has been found with other species, behavior on the concurrent schedules was biased by the type of feed, with the 6% salt and the coconut giving the greatest biases towards the barley-carob mixture. The schedules were varied over 17 conditions using the barley-carob mixture alone or the barley-carob mixture versus the mixture with 4% or 6% salt. Both the contingency-discriminability model (Davison & Jenkins, 1985) and the generalized matching law described the data from the three sets of conditions equally well. Both gave similar measures of bias; however, some of the parameter values found with the contingency discriminability model were uninterpretable. Thus, any argument for this model based on the interpretability of the parameter values becomes weak. It is worth retaining the generalized matching law as a descriptor of such data.
Journal of the experimental analysis of behavior, 2003 · doi:10.1901/jeab.2003.79-289