Too busy to mind: neural mechanisms driving future remembering in a working memory task

The mechanisms bridging perception, working memory (WM) and long-term memory are poorly understood. Previous evidence showed that high vs. low WM cognitive load deteriorates long-term encoding of the presented stimuli (Axmacher et al., 2009) and parametrically engages high-order visual areas selective for the stimulus content (Druzgal et al., 2001). However, studies attempting to integrate perception and information transfer between different memory storages are missing. We exploited the Human Connectome Project 1200 Subjects Release dataset (Van Essen et al., 2013) by analyzing functional magnetic resonance imaging (fMRI) and behavioral data of an N-back paradigm. We sought to dissociate the neural underpinnings of remembering or forgetting stimuli belonging to different categories after their encoding in short-term (0-back) or working (2-back) memory. Methods During the fMRI exam, participants performed a 0- and a 2-back task on images of faces and scenes. We distinguished between target and lure (i.e., distractors) stimuli. In the 0-back task, the first image of the sequence acted as cue stimulus, which the following images could match (targets) or not, eventually being repeatedly presented (lures). In the 2-back task, 2-back repeats were the targets, 1- and 3-back repeats were considered as lures. In a subsequent behavioral experiment, subjects were asked to recognize the previously seen visual stimuli among distractors to test their long-term trace. We performed univariate analyses on fMRI data (N = 165) and mixed effect models on the behavioral recognition scores (N = 1026) modeling the trial type (target or lure) and the category (face or place) as fixed factors, separately for the 0- and the 2-back task to disambiguate task-specific mechanisms. Results The 0-back task was easier in terms of accuracy rates during the fMRI tasks and yielded higher long-term recognition scores in the post-scan behavioral task. Recognition was affected by category and trial type for both the 0- and 2-back task: faces and targets were more likely to be recognized than scenes and lures, respectively. Future remembering was associated with higher activity, for each category, in the respective category-specific areas (occipital face area and retrosplenial cortex) in the 0-back task. Future remembering also interacted with trial type: remembered targets (vs. lures) yielded higher activity in the bilateral temporo-occipital areas and the right superior parietal lobule in the 0-back task, and lower activity in the left dorsolateral prefrontal cortex (dlPFC) in the 2-back task. Conclusion With the present study, we unveiled the mechanisms bridging perception and higher-level memory functions. We showed that both bottom-up (i.e., stimulus content) and top-down (i.e., task complexity) features affect the successful encoding of stimuli into the long-term storage. The involvement of partially segregated neural networks accounts for task-dependent effects on future recognition. When compared to lures, 0-back targets enhanced the recruitment of perceptual and attentional networks that ensured the consolidation of such stimuli in the long-term memory. When facing the 2-back lure trials, the dlPFC as part of the Central Executive Network precluded fast and automatic responses. The cognitive effort necessary to solve such highly conflicting trials weakened the long-term encoding of lure stimuli. Together, we show that the consolidation of information temporarily stored in the short-term and working memory storages is fostered by largely independent neural networks. References Axmacher, N., Haupt, S., Cohen, M. X., Elger, C. E., & Fell, J. (2009). Interference of working memory load with long‐term memory formation. European Journal of Neuroscience, 29(7), 1501-1513. Druzgal, T. J., & D’Esposito, M. (2001). Activity in fusiform face area modulated as a function of working memory load. Cognitive Brain Research, 10(3), 355-364. Van Essen, D. C., Smith, S. M., Barch, D. M., Behrens, T. E., Yacoub, E., Ugurbil, K., & Wu-Minn HCP Consortium. (2013). The WU-Minn human connectome project: an overview. Neuroimage, 80, 62-79.