Boran, Ece, et al. Science Advances 5.3 (2019).
Review prepared by: Martina Maier
The paper aims to clarify the involvement of the hippocampus in working memory and how it interacts with the cortex. They further hypothesized that hippocampal neurons will show a neuronal firing pattern that is dependent on memory load. In this vein, 9 subjects that had depth electrodes implanted in the medial temporal lope for later surgical treatment of epilepsy, participated in a working memory task while intracranial neural activity (single neuron and iEEG) and EEG were recorded. The memory load in the task, a modified version of the Sternberg WM task, was systematically modulated by having the subject to remember either four, six or eight letters. According to behavioral data, all the subjects mastered the task, and the memory load had the expected decreasing effect on accuracy and an increasing effect on reaction time. However, it must be said, that the subject could have been overtrained, as some of them performed the task up to seven times. Analysis of the neuronal data revealed that 20.6% of recorded hippocampal neurons showed higher firing rate during maintenance than during pre-stimulus presentation, which is significantly more neurons than in entorhinal cortex or amygdala, and the firing rate increased with the workload. In contrast, the neurons in entorhinal cortex showed increased firing only when the probe for recall was shown. Using a support vector machine, it could be further confirmed that the firing of the hippocampal neurons during the maintenance phase could predict memory load although only between low and high. The firing of the same neurons the encoding phase was predictive for whether the subject would respond right or wrong. Analysis of the EEG data pointed to an increase in power during the maintenance phase for high memory load in the alpha band (8-12 Hz) in the parietal cortex (PZ electrode), which was also found in iEEG and neuronal firing of the hippocampus. Further calculation of the phase-locking value (PLV) pointed to synchronization between i EEG of hippocampus and EEG of parietal cortex in the alpha band. However, these findings of alpha band increase (EEG and iEEG) and the synchronization was only detailed out for 1 subject. For the other eight subjects, only PLV was computed and pointed to similar results (e.g. significance in the alpha band, some memory load dependence and increasing during maintenance for hippocampus iEEG and PZ EEG), however, subject variability was high. This is most probably the reason no group effect could be shown. So although the authors have found some evidence of hippocampal involvement in working memory functioning (firing was predictive for success or failure, and differentiated between high and low load during maintenance period), evidence for cortico-hippocampal coupling is weak and requires further investigation.