Comparing learned predictiveness effects within and across compound discriminations.

Area of Science:

• Cognitive Psychology
• Learning and Memory
• Behavioral Science

Background:

• Learned predictiveness influences how organisms learn about their environment.
• Previous theories suggested that comparing predictors is crucial for learning their value.

Purpose of the Study:

• To investigate whether learned predictiveness relies on direct comparisons between predictors.
• To determine if the associability of a cue changes based on its relative predictive accuracy.

Main Methods:

• Four human learning experiments were conducted.
• Participants completed initial discrimination tasks (linear compound, biconditional, pseudodiscrimination).
• Subsequent learning and testing stages measured changes in stimulus associability.

Main Results:

• Learned predictiveness effects were observed across all experiments.
• Changes in stimulus associability were consistent regardless of the presence of equally, more, or less predictive stimuli.
• No evidence was found that relative comparison between cues controls learned associability.

Conclusions:

• Learned associability is determined by the absolute predictiveness of individual cues.
• Competitive allocation of attention between cues does not appear to be the primary mechanism.
• Findings challenge theories emphasizing relative comparison in learned predictiveness.