Surprise and destabilize: prediction error influences episodic memory reconsolidation

Generalising reconsolidation: Spatial context and prediction error

Activating a previously consolidated memory trace brings it back into a labile state where it must then undergo a re-stabilisation process known as reconsolidation. During this process memories are susceptible to interference and may be updated with new information. In the studies showing this effect in human episodic memory, the reconsolidation process has been triggered primarily using spatial context or prediction error manipulations to reactivate an established memory. However, these studies have produced conflicting results, showing both that spatial context is necessary and sufficient to trigger reconsolidation and that prediction error is necessary and sufficient to trigger the process. We examined this conflict in two experiments, one investigating the role of context cues and another investigating the role of prediction error. In Experiment 1, spatial context triggered a reconsolidation process and prediction error was irrelevant. In Experiment 2, prediction error triggered reconsolidation, and spatial context cues were irrelevant. These findings replicate prior research but add to the puzzle concerning the roles of these two means of triggering reconsolidation.

Reward Prediction Error and Declarative Memory

Learning based on reward prediction error (RPE) was originally proposed in the context of nondeclarative memory. We postulate that RPE may support declarative memory as well. Indeed, recent years have witnessed a number of independent empirical studies reporting effects of RPE on declarative memory. We provide a brief overview of these studies, identify emerging patterns, and discuss open issues such as the role of signed versus unsigned RPEs in declarative learning.

Examining the Use of Temporal-Difference Incremental Delta-Bar-Delta for Real-World Predictive Knowledge Architectures

Predictions and predictive knowledge have seen recent success in improving not only robot control but also other applications ranging from industrial process control to rehabilitation. A property that makes these predictive approaches well-suited for robotics is that they can be learned online and incrementally through interaction with the environment. However, a remaining challenge for many prediction-learning approaches is an appropriate choice of prediction-learning parameters, especially parameters that control the magnitude of a learning machine's updates to its predictions (the learning rates or step sizes). Typically, these parameters are chosen based on an extensive parameter search—an approach that neither scales well nor is well-suited for tasks that require changing step sizes due to non-stationarity. To begin to address this challenge, we examine the use of online step-size adaptation using the Modular Prosthetic Limb: a sensor-rich robotic arm intended for use by persons with amputations. Our method of choice, Temporal-Difference Incremental Delta-Bar-Delta (TIDBD), learns and adapts step sizes on a feature level; importantly, TIDBD allows step-size tuning and representation learning to occur at the same time. As a first contribution, we show that TIDBD is a practical alternative for classic Temporal-Difference (TD) learning via an extensive parameter search. Both approaches perform comparably in terms of predicting future aspects of a robotic data stream, but TD only achieves comparable performance with a carefully hand-tuned learning rate, while TIDBD uses a robust meta-parameter and tunes its own learning rates. Secondly, our results show that for this particular application TIDBD allows the system to automatically detect patterns characteristic of sensor failures common to a number of robotic applications. As a third contribution, we investigate the sensitivity of classic TD and TIDBD with respect to the initial step-size values on our robotic data set, reaffirming the robustness of TIDBD as shown in previous papers. Together, these results promise to improve the ability of robotic devices to learn from interactions with their environments in a robust way, providing key capabilities for autonomous agents and robots.

Effects of transcranial direct current stimulation over the posterior parietal cortex on episodic memory reconsolidation

Consolidated memories may return to labile/unstable states after their reactivation, thus requiring a restabilization process that is known as reconsolidation. During this time-limited reconsolidation window, reactivated existing memories can be strengthened, weakened or updated with new information. Previous studies have shown that non-invasive stimulation of the lateral prefrontal cortex after memory reactivation strengthened existing verbal episodic memories through reconsolidation, an effect documented by enhanced delayed memory recall (24 h post-reactivation). However, it remains unknown whether the left posterior parietal cortex (PPC), a region involved during reactivation of existing episodic memories, contributes to reconsolidation. To address this question, in this double-blind experiment healthy participants (n = 27) received transcranial direct current stimulation (tDCS) with the anode over the left PPC after reactivation of previously learned verbal episodic memories. Memory recall was tested 24 h later. To rule out unspecific effects of memory reactivation or tDCS alone, we included two control groups: one that receives tDCS with the anode over the left PPC without reactivation (n = 27) and another one that receives tDCS with the anode over a control site (primary visual cortex) after reactivation (n = 27). We hypothesized that tDCS with the anode over the left PPC after memory reactivation would enhance delayed recall through reconsolidation relative to the two control groups. No significant between groups differences in the mean number of words recalled on day 3 occurred, suggesting no beneficial effect of tDCS over the left PPC. Alternative explanations were discussed, including efficacy of tDCS, different stimulation parameters, electrode montage, and stimulation site within the PPC.

Retrieval of retrained and reconsolidated memories are associated with a distinct neural network

Consolidated memories can persist from a single day to years, and persistence is improved by retraining or retrieval-mediated plasticity. One retrieval-based way to strengthen memory is the reconsolidation process. Strengthening occurs simply by the presentation of specific cues associated with the original learning. This enhancement function has a fundamental role in the maintenance of memory relevance in animals everyday life. In the present study, we made a step forward in the identification of brain correlates imprinted by the reconsolidation process studying the long-term neural consequences when the strengthened memory is stable again. To reach such a goal, we compared the retention of paired-associate memories that went through retraining process or were labilizated-reconsolidated. Using functional magnetic resonance imaging (fMRI), we studied the specific areas activated during retrieval and analyzed the functional connectivity of the whole brain associated with the event-related design. We used Graph Theory tools to analyze the global features of the network. We show that reconsolidated memories imprint a more locally efficient network that is better at exchanging information, compared with memories that were retrained or untreated. For the first time, we report a method to elucidate the neural footprints associated with a relevant function of memory reconsolidation.