The effectiveness of targeted memory reactivation in modulating attentional bias

Attention is a critical cognitive function that helps us filter and focus on relevant stimuli. Emotions play a crucial role in shaping our attention and steering our daily decisions and actions. Attentional bias (AB) in clinical populations refers to the tendency to focus on negative information while ignoring or discounting positive or neutral stimuli. This scenario leads to distorted cognitive processes, contributing to the development and maintenance of psychopathology. An AB index is provided by reaction times (RTs) in the dot-probe task (DPT), which measures the time taken to direct attention towards a target stimulus (i.e., a white dot) appearing after the simultaneous presentation of two emotional stimuli (negative/positive). Individuals with pathological AB show quicker responses and larger P300 amplitudes to dots replacing negative images than positive ones. Attentional Bias Modification (ABM) aims to correct these biases by training individuals to redirect their attention to positive stimuli. By increasing the probability of dot appearance to 80% behind positive images, individuals implicitly learn to redirect their attention away from negative stimuli and towards positive/neutral stimuli. Through repeated training sessions, ABM can reduce AB towards negative stimuli by leveraging the brain's natural learning and implicit memory consolidation processes and, presumably, creating lasting changes in attentional patterns. Memory consolidation is critical for stabilizing and enhancing memories after initial acquisition, with sleep playing a crucial role. In this context, the Targeted Memory Reactivation (TMR) paradigm is an experimental technique that strengthens specific memory traces during sleep through sensory stimuli (odors/sounds) delivered primarily during slow-wave sleep (SWS). Our study investigates the effect of TMR on modulating AB at behavioral and electrophysiological levels. Twenty-two female students (mean age ± standard deviation, 21.86 years ± 2.64; age range: 20-30 years) were involved in a between-subjects study. AB was measured using a modified version of DPT, along with high-density EEG, before (T1) and after (T2) a nocturnal polysomnography with Acoustic Stimulation (ASTMR vs. ASCONTROL) during SWS. During the DPT, participants viewed pairs of emotional faces (happy/sad) followed by a dot that replaced the position previously occupied by the happy (Positive Valence trial) or sad faces (Negative Valence trial), appearing 50% of the time in each position. Participants were required to quickly and accurately indicate the dot’s position (left vs. right) using the keyboard. The participant’s response triggered a different auditory cue for Positive Valence trials and for Negative Valence trials. To ensure the explicit learning of the association of each sound with the specific emotional valence of the images, so allowing the effective reactivation of the corresponding memory traces during TMR, participants underwent the Association Task during the T1 session, after the DPT. This task involved presenting the 200 image pairs used in the DPT along with one of the two sounds associated with positive or negative valence trial. Participants had to drag a dot from the center of the screen to the emotional face with the valence corresponding to the sound presented. During SWS, the ASTMR group was exposed to the sound stimuli linked to the Positive Valence trials, while the ASCONTROL group heard a novel sound. To evaluate the effects of TMR, Mixed Model Analysis was applied to RTs and P300 amplitude and latency of the event-related potential (ERP) generated in response to the dot, including valence (Positive Valence vs. Negative Valence), session (T1 vs. T2), and group (ASTMR vs. ASCONTROL) as within/between factors. The EEG correlates of TMR were assessed by comparing the amplitude of ERPs and the event-related spectral power perturbation (ERSP) time-locked to the acoustic stimulation between the ASTMR and ASCONTROL groups (unpaired t-tests). Behaviorally, TMR led to a greater reduction in RTs to the DPT in the ASTMR group compared to ASCONTROL that was not valence-specific, indicating a general improvement in processing speed without effect on the AB. Instead, at the electrophysiological level, TMR had the expected valence-specific effects on P300 in response to the dot. Specifically, ASTMR group showed a significant reduction in P300 amplitude and latency in the (reactivated) positive valence trials only, which was not observed in the ASCONTROL group. Additionally, the EEG correlates of TMR during sleep compared to those linked to the presentation of a completely new sound indicated a cortical response characterized by higher amplitude in the ERP within the 750-1200 ms interval post-stimulation associated with increased spindle activity in the ERSP analysis, suggesting an effective memory reactivation leading to successful memory consolidation in the ASTMR group. The results indicate that TMR did not significantly alter AB behaviorally in healthy subjects. However, it demonstrated potential to effectively modulate electrophysiological responses to emotional stimuli, facilitating the processing of positive valence ones. Future studies should explore the effects of TMR in clinical populations, where the impact is expected to be more pronounced due to the presence of maladaptive AB, potentially affecting both EEG and behavioral measures. Additionally, investigating repeated TMR sessions using home-based EEG technologies in more naturalistic settings could provide a promising avenue for interventions targeting pathological AB. This scenario could open new avenues for developing sleep-based approaches to modify AB and complement existing therapies for various psychological disorders.