Alpha-band Amplitude During Memory Encoding Is an Index of Memory Performance

Abstract

Alpha-band (8–13 Hz) rhythm during encoding of sequentially presenting memory items was recorded using a 76-channel magnetoencephalograph. In a memory task, seven images each with an arrow pointing up, down, left or right were presented sequentially. Thereafter, a recall number was presented. Each participant memorized the arrow directions in the order of presentation and recalled the direction of the arrow for the recall number, and responded by pressing a button. Each partici- pant also took part in a control task, in which the participant looked at the same images as those in the memory task but pressed the button previously instructed. Ten right-handed male volunteers performed 280 epochs of the memory task and 120 epochs of control task. The amplitudes of alpha-band rhythm during presentation of midterm (3rd–5th) items were compared with those of beginning (1st and 2nd) items. Fifteen sensors showed significant differences (midterm > beginning) in the memory task (0.01 < p < 0.05 at 11 sensors, 0.001 < p < 0.01 at 4 sensors), while no sensors showed a significant increase in the control task. Using the statistical parametric mapping (SPM) method, the region of increase in alpha-band rhythm from beginning to midterm was estimated to be located in the visual cortex and around bilateral temporal gyri. Additionally, the increase in amplitude from be- ginning to midterm was significantly greater before a correct answer than before a wrong answer (0.01 < p < 0.05 at 7 sensors and p < 0.001 at 1 sensor). Previous works demonstrated that alpha-band rhythm is increased by active inhibition of visual in- puts that are irrelevant to memory. The region of increase estimated to be in the visual cortex agrees with previous works. An in- terpretation of our results is that since irrelevant visual inputs are successfully inhibited before a correct answer, the amplitude of alpha-band rhythm increases more than that before a wrong answer. These results indicate the possibility that alpha-band am- plitude during memory encoding can be used as an index of memory performance