Saying and doing: A contingency-space analysis.
Check the four say-do boxes to see why a client’s promise fails, then reinforce the action, not the words.
01Research in Context
What this study did
Matthews et al. (1987) drew a four-box map.
One axis: does the child say the right thing?
Other axis: does the child do the right thing?
The map shows four kinds of matches and mismatches.
No kids were tested; the paper is a blueprint.
What they found
The map gives names to every say-do pair.
Example: child promises to share and then shares = true promise.
Child promises and does not share = false promise.
Having names lets you pick the right fix for each mismatch.
How this fits with other research
Wilson et al. (1987) slammed the map the same year.
They said, "Just count if the words and actions match."
The fight looks big, but both want accurate data.
Iwata et al. (1990) later tested preschoolers.
They proved kids only keep promises when reinforcement waits at the end.
The map turned into real procedures: reward the match, not the talk.
Périkel et al. (1974) did the math first; Matthews et al. (1987) moved it to kid promises.
Why it matters
Use the four-box check before you write a plan.
If a client says he will finish tasks but does not, you have a false promise.
Add a reinforcer for the doing, not the saying.
Skip rewards for empty talk and watch real work show up.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →After a client states a plan, mark say-do match or mismatch on your data sheet and deliver praise only if both line up.
02At a glance
03Original abstract
Correspondences between verbal responding (saying) and nonverbal responding (doing) may be organized in terms of the classes of verbal/nonverbal relations into which particular instances of verbal/nonverbal response sequences can enter. Contingency spaces, which display relations among events in terms of the probability of one event given or not given another, have been useful in analyses of nonverbal behavior. We derive a taxonomy of verbal/nonverbal behavior relations from a contingency space that takes into account two conditional probabilities: the probability of a nonverbal response given a verbal response and that probability given the absence of the verbal response. For example, positive correspondence may be said to exist as a response class when the probability of doing is high given saying but is otherwise low. Criteria for other generalized classes, including negative correspondence, follow from this analysis.
Journal of applied behavior analysis, 1987 · doi:10.1901/jaba.1987.20-69