Does a signal help or hurt learning that my action caused an outcome?
A brief signal between an action and its outcome hurt learning when the outcome came quickly but helped it when the outcome was delayed.
Source
Temporal contiguity determines overshadowing and potentiation of human Action-Outcome performance
Study at a glance
- Design
- Human experiment — Three free-operant computer-game experiments manipulating action-outcome delay, signal timing and signal salience within subjects
- N
- N=81 · 81 undergraduates in Experiment 1 (30 in lab, 51 online), 81 in Experiment 2, 82 in Experiment 3
- Population
- Undergraduate students in the UK and Spain
- Outcome
- Rate of space-bar pressing (instrumental performance) and 0-100 causal judgments
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
Students played a game where pressing the space bar caused explosions after a 2-second or 6-second delay, sometimes with a brief grey sky flash (the signal) in between. They measured how often people pressed and how strongly they believed their pressing caused the explosions. Experiment 2 moved the signal to the start or end of a 6-second delay, and Experiment 3 made the signal dimmer or brighter under a 2-second delay.
What they found
In Experiment 1 the signal reduced pressing with the 2-second delay (overshadowing) but increased pressing and causal ratings with the 6-second delay (potentiation). In Experiment 2 the signal only helped when it came right after the action, not when it came just before the outcome. In Experiment 3 a bright, salient signal reduced pressing more than a dim one, but causal ratings were unaffected.
The limits
What it doesn't show
The samples were mostly female undergraduates, and part of Experiment 1 was run online on participants' own computers, so control over the setting was limited. Causal judgments often did not follow the pressing data, so the two measures seem to dissociate and the effect on beliefs is weaker than on behaviour. The signal was ambiguous (it could mean a comrade was shooting) and only ratio schedules were used, which the authors say may have favoured facilitation; interval schedules or clearer instructions might give different results.
Key terms
- Overshadowing
- When a second cue that also predicts an outcome reduces how much is learned about the first cue.
- Potentiation
- The opposite of overshadowing: an added cue increases learning about the target cue or action.
- Temporal contiguity
- How close in time two events are; here, the gap between pressing and the explosion.
- Free-operant procedure
- A task where participants can respond as often as they like, and response rate is the measure of learning.
- Configural learning
- Pearce's idea that a compound of stimuli is learned as a single whole unit rather than as separate elements.
Flashcards
0 of 10 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
With a 2-second action-outcome delay, the intervening signal mainly:
Common questions
Why would standard theories predict the signal always hurts learning?
Models like Rescorla-Wagner assume cues that predict the same outcome share a limited amount of associative strength, so a signal should always take some away from the action.
How do the authors explain the facilitation with long delays?
They propose that with a long delay the unique features of the action and signal fade, so the two are encoded as one configuration and learning about that compound generalises to the action alone.
Was the effect on beliefs as strong as on behaviour?
No. With a short delay the signal lowered pressing but did not change causal ratings, and the salience manipulation in Experiment 3 left ratings unchanged.
More on Learning