Reward conditioning may not have an effect on category-specific memory

Priyanka Sukumaran, Nina Kazanina, Conor Houghton

Abstract

Behavioural tagging facilitates the temporary storage of seemingly insignificant episodic events, which may later become salient and enhanced in memory. Human behavioural studies have demonstrated selective memory enhancement for neutral stimuli from one category when this category is subsequently paired with reward. Although this phenomenon has implications for the role of reward conditioning on emotional and adaptive memory, its generalisability is underexplored. We conducted four experiments to investigate whether pairing items from a semantic category, animals or objects, with high or low rewards resulted in preferential memory for the high-reward category. Three of these experiments also aimed to replicate the category-specific retrospective enhancement effect reported by Patil et al. and two explored the corresponding prospective memory effect. None of our experiments showed consistent evidence for an effect of reward on category-specific memory enhancement, despite employing the same reward paradigm and incidental encoding protocol as in the original study. Consequently, we found no evidence for category-specific retrospective or prospective enhancement effects. Our experiments were conducted online which is an equally relevant method for assessing behavioural phenomenon as the in-person studies conducted by Patil et al. Overall, our results question the generalisability of previously reported category-specific memory enhancement effects due to reward.

Introduction

In neuroscience, the synaptic tag-and-capture hypothesis suggests that our brains "tag" seemingly insignificant events. If a salient event, such as a reward or a shock, occurs shortly after, these tags are "captured," leading to memory consolidation. A high-profile 2017 study by Patil et al. suggested that this effect is category-specific: rewarding new images from a specific category (e.g., animals) supposedly boosts memory for previously seen neutral items from that same category. This "Reward Retroactively Enhances Memory" (RREM) effect implies that adaptive memory systems can selectively rescue past information based on semantic relationships. However, replication difficulties have led to questions regarding the reliability and generalizability of this phenomenon.

Methods

The researchers conducted four high-powered online experiments (N=410 total) to investigate the robustness of the RREM effect. The experimental design followed a structured timeline involving three main phases:

  1. Incidental Encoding (Pre-Conditioning): Participants performed tasks, such as matching images or learning Japanese word-image pairs, without knowledge of a subsequent memory test.
  2. Reward Conditioning: One semantic category (e.g., animals) was associated with a "high reward" (£0.15 or a £5 bonus), while another (e.g., objects) was associated with a "low reward" (£0.01 or £0.25 bonus).
  3. Post-Conditioning Phase: Additional items were presented to test for prospective memory effects.
Experimental timeline and task phases

A surprise recognition test was administered 24 hours later. Memory performance was quantified using Corrected Recognition (Hits minus False Alarms) and d-prime ($d'$), a signal detection measure that accounts for response bias. Stimuli included the original set from Patil et al. (240 images of animals and tools) and 384 Japanese nouns for linguistic testing in Experiment 1.

Key Results

Across all four experiments, the researchers found no consistent evidence for category-specific memory enhancement. While some measures of "Corrected Recognition" showed minor effects, the more stringent $d'$ measure—which filters out the tendency to guess "old" for rewarded categories—did not support the RREM hypothesis.

Memory performance (d-prime) across experimental phases

Bayesian analysis further supported the null hypothesis, with most Bayes Factors ($BF_{10}$) falling below 0.33, indicating substantial evidence that reward conditioning had no effect on category-specific memory. Interestingly, while the category-specific boost failed, items presented during the reward phase were remembered better than those in non-reward phases, confirming that reward does enhance memory locally, even if it does not "spread" to related past or future events.

Discussion and Conclusion

The failure to replicate the RREM effect suggests that the phenomenon may be less robust than previously reported. Several factors might explain the discrepancy, including the magnitude of the reward (the original study used higher stakes) and the level of task engagement. The authors suggest that active semantic processing, such as explicitly naming categories, might be necessary to link rewards to semantic tags.

Furthermore, the study highlights the importance of using signal detection measures like $d'$ to distinguish between actual memory sensitivity and simple response bias. While the idea of "retroactive memory rescue" remains a compelling theoretical bridge between cellular biology and human behavior, these findings urge caution in assuming its broad applicability to everyday semantic categories.

Summary updated

Abstract

Behavioural tagging facilitates the temporary storage of seemingly insignificant episodic events, which may later become salient and enhanced in memory. Human behavioural studies have demonstrated selective memory enhancement for neutral stimuli from one category when this category is subsequently paired with reward. Although this phenomenon has implications for the role of reward conditioning on emotional and adaptive memory, its generalisability is underexplored. We conducted four experiments to investigate whether pairing items from a semantic category, animals or objects, with high or low rewards resulted in preferential memory for the high-reward category. Three of these experiments also aimed to replicate the category-specific retrospective enhancement effect reported by Patil et al. and two explored the corresponding prospective memory effect. None of our experiments showed consistent evidence for an effect of reward on category-specific memory enhancement, despite employing the same reward paradigm and incidental encoding protocol as in the original study. Consequently, we found no evidence for category-specific retrospective or prospective enhancement effects. Our experiments were conducted online which is an equally relevant method for assessing behavioural phenomenon as the in-person studies conducted by Patil et al. Overall, our results question the generalisability of previously reported category-specific memory enhancement effects due to reward.
BibTeX
@inproceedings{sukumaran2023reward,
  author = {Priyanka Sukumaran and Nina Kazanina and Conor Houghton},
  title = {{Reward conditioning may not have an effect on category-specific memory}},
  booktitle = {Scientific},
  year = {2023},
  doi = {10.1038/s41598-023-48874-z},
}
Figure