Summary: Unexpectedly positive outcomes strengthen memory for specific events. These results may have implications for understanding and treating memory problems associated with depression.
Source: Brown University
Key finding: Researchers found that moments when outcomes exceed expectations — for example, finding an unexpectedly good parking spot or discovering a $20 bill on the sidewalk — not only influence generalized or integrated memories but also enhance memory for incidental, event-specific details.
To illustrate, remembering exactly where you parked this morning is an example of a specific episodic memory, while recalling generally reliable parking spots across trips is an integrated memory. The new work shows that the incidental details tied to a single event — such as whether the tree beside your car was a spruce or a maple — can be reinforced when the outcome of that event is better than anticipated.
“Our new finding is that incidental, irrelevant details from specific events are strengthened by unexpectedly good outcomes,” said Matt Nassar, assistant professor of neuroscience at Brown University and the study’s corresponding author. He noted that these effects might be especially relevant for people with depression, who often recall negative experiences more readily than positive ones. The research team aims to explore potential therapeutic applications, but cautioned that clinical treatments are not yet developed.
Mechanistically, the phenomenon is linked to reward-prediction error — the difference between expected and actual outcomes — which is associated with dopamine release in brain systems that support memory. Since depression has been associated with imbalances in neurotransmitters including dopamine, serotonin and norepinephrine, the new results suggest a route by which depressed individuals may fail to form strong positive memories, contributing to a negative memory bias.
Daniel Dillon, a co-author and researcher at McLean Hospital and Harvard Medical School, emphasizes that investigating the clinical interaction between depression and memory is a primary goal of his work. The study was published on May 6 in the journal Nature Human Behaviour.
The experiment used a two-phase design: a learning phase followed by a memory test. During the learning task, across 160 rounds participants saw a point value (between 1 and 100), were shown an image of either a living or an inanimate object, and chose whether to gamble on a coin-flip outcome or pass. If they gambled and lost, they lost 10 points; if they gambled and won, they earned the round’s point value. At the end of the session, accumulated points were converted into a small monetary payout, typically under $5.
Participants were not told that the specific images would later be tested. They were informed only that the object category (living versus inanimate) was tied to their chance of winning. After the learning phase, the researchers tested participants’ memory for the precise images. The team found that memory for a specific image was improved when that image appeared in a round associated with a high probability of winning and when the participant experienced a positive reward-prediction error at the time the image was shown. In other words, stronger memory encoding occurred when the positive surprise coincided with the presentation of the item.
The study included more than 250 participants recruited through Amazon Mechanical Turk. Some participants paid close attention to the value and probability information during the gambling rounds, while others responded more randomly. Those who attended more closely to the task demonstrated stronger memory performance, and the researchers used these individual differences to refine a computational model of how memory strength related to reward-prediction errors.
Surprisingly, the enhancement in memory encoding tied to positive prediction errors appeared quickly: memory for high-value rounds that resulted in wins was equally robust when tested five minutes after learning and when tested 24 hours later. This indicates that, for this task, the effect did not strongly depend on longer-term consolidation processes. The authors note that this finding differs from some earlier animal studies that emphasized a delayed consolidation role for dopamine, and they plan follow-up work to reconcile these timing differences across species and experimental designs.
The team intends to extend this research by rerunning the experiment with people diagnosed with depression to determine whether the same reward-related memory enhancements occur. Other planned directions include mapping the neural circuits that connect reward-prediction error signals to episodic memory encoding and clarifying how episodic memory systems interact with systems that form integrated memories over longer timescales.
Funding: This research was supported by the National Institutes of Health (grants F32MH102009, K99AG054732 and R01 MH080066) and the National Science Foundation (grant 1460604).
Source:
Brown University
Media contact:
Mollie Rappe – Brown University
Original research: The paper, titled “Positive reward prediction errors during decision-making strengthen memory encoding,” was authored by Anthony I. Jang, Matthew R. Nassar, Daniel G. Dillon and Michael J. Frank and published in Nature Human Behaviour. The study describes an incidental memory paradigm showing that reward prediction errors enhance episodic encoding when the timing of the positive surprise aligns with the presentation of memory items; this enhancement was detectable within minutes and did not differ substantially after 24 hours.