The Effect of Prediction Error on Episodic Memory Encoding is Modulated by the Outcome of the Predictions

Abstract

Predictive processing accounts propose that our brain constantly tries to match top-down internal representations with bottom-up incoming information from the environment. Predictions can lead to prediction errors of varying degrees depending on the extent to which the information encountered in the environment conforms with prior expectations. Theoretical and computational models assume that prediction errors have beneficial effects on learning and memory. However, while there is strong evidence on the effects of prediction error on learning, relatively less evidence is available regarding its effects on memory. Moreover, most of the studies available so far manipulated prediction error by using monetary rewards, whereas in everyday life learning does not always occur in the presence of explicit rewards. We used a task in which participants leaned context/object-category associations of different strength based on the outcomes of their predictions. After learning these associations, participants were presented with trial-unique objects that could match or violate their predictions. Finally, participants were asked to complete a surprise recognition memory test. We used a reinforcement learning model to derive subject-specific trial-to-trial estimates of prediction error at encoding and link it to subsequent recognition memory. Results showed that model-derived prediction errors at encoding influenced subsequent memory as a function of the outcome of participants’ predictions (correct vs incorrect). When participants correctly predicted the object category, stronger prediction errors (as a consequence of weak expectations) led to enhanced memory. In contrast, when participants incorrectly predicted the object category, stronger prediction errors (as a consequence of strong expectations) led to impaired memory. These results reveal a computationally specific influence of prediction error on memory formation, highlighting the important moderating role of choice outcome that may be related to interactions between the hippocampal and striatal dopaminergic systems.

Signed and Unsigned Effects of Prediction Error on Memory: Is it a Matter of Choice?

“…Studies taking this approach generally found a positive relation between PE and memory (Calderon et al, 2021;Davidow et al, 2016;De Loof et al, 2018;Jang et al, 2019;Pupillo et al, 2023;Rosenbaum et al, 2022;Niv, 2019, 2021;Rouhani et al, 2018). This PE-driven enhancement of recognition memory does not depend on specific characteristics of the task, such as whether the values of the stimuli were learned before or during the encoding task, whether the outcome was monetary or not, and whether the recognition test was intentional or incidental.…”

“…However, regarding PE effects at outcome, the findings diverge (Ergo et al, 2020). In fact, while some studies have found unsigned PEs effects at outcome (Rouhani and Niv, 2021;Rouhani et al, 2018), others have reported outcome-related signed PEs effects (Calderon et al, 2021;De Loof et al, 2018;Pupillo et al, 2023).…”

“…Although there is mounting evidence that PEs positively affect episodic memory, the findings diverge when it comes to the sign of the PE. On the one hand, a series of studies found an effect of signed PEs on recognition memory (Calderon et al, 2021;Davidow et al, 2016;De Loof et al, 2018;Jang et al, 2019;Pupillo et al, 2023;Rouhani and Niv, 2021). That is, these studies suggest that objects related to better-than-expected outcomes (positive PE) are remembered better than objects related to worse-than-expected outcomes (negative PE; Fig.…”

“…These models have frequently been used in the learning literature and show a marked correspondence with behavioural and neural data (Daw and Tobler, 2014). Moreover, RL algorithms have been used in the majority of studies on the relationship between PE and episodic memory (Calderon et al, 2021;Davidow et al, 2016;De Loof et al, 2018;Pupillo et al, 2023;Rouhani and Niv, 2021;Rouhani et al, 2018;Wimmer et al, 2014), and we therefore also focus on an RL approach in this work.…”

“…A major critical open question concerns the sign of the effect of PEs (Ergo et al, 2020). While several studies showed a positive relationship between signed PEs (i.e., ranging from negative to positive) and episodic encoding (Calderon et al, 2021;Davidow et al, 2016;De Loof et al, 2018;Jang et al, 2019;Pupillo et al, 2023), other studies reported a positive effect of unsigned PEs (i.e., absolute PEs; Niv, 2019, 2021;Rouhani et al, 2018). A positive relationship between signed PEs and memory suggests that episodic encoding is better for better-than-expected outcomes compared to worse-than-expected outcomes.…”