Predictability Changes What We Remember in Familiar Temporal Contexts.

Area of Science:

• Cognitive Neuroscience
• Neurobiology of Memory

Background:

• The human brain uses past memories to predict future events, enhancing responsiveness.
• Prediction errors can lead to memory pruning and forgetting, weakening neural representations.

Purpose of the Study:

• To investigate if memories are protected from pruning when abstract predictions are accurate, despite item-level prediction errors.
• To understand how the reliability of categorical predictions influences memory updating.

Main Methods:

• Functional magnetic resonance imaging (fMRI) pattern classification was employed.
• Participants viewed sequences of object cues followed by novel items.
• Experimental conditions manipulated the predictability of item categories following cues.

Main Results:

• Memory pruning occurred only in contexts with unpredictable item categories.
• Encoding of new items was less effective in predictable contexts.
• Category-specific predictions were identified using fMRI data.

Conclusions:

• Associative memory updating is modulated by the reliability of abstract-level predictions.
• The brain selectively prunes memories based on prediction error, but spares them when higher-level predictions are consistent.
• Contextual predictability influences both memory pruning and the encoding of new information.