The search for correlates of memory in the EEG has been a confounding area of study. The answer may lie in clever approaches to temporal structure and synchronization across sensory modalities.
Although we may subjectively think that our memories play out in our minds much like a movie reel, our episodic memories – memories of specific events constructed of various modes of sensory stimulii – are actually constructed and reconstructed representations of a particular experience. These constructions are composed of associations of place, time and sensations which are forged together by a network of memory regions in our brain which includes our hippocampus.
A complicated picture
But if you asked researchers what aspect of the EEG was associated with memory, you might get a multitude of different answers. That’s because nearly every EEG frequency band has been implicated in memory in some way, with both increases and decreases in amplitude/power reported. This confounding picture of EEG frequency bands has left researchers struggling to understand the “big picture”.
See related post The Blue Frog in the EEG
The seminal work of Donald Hebb in 1949 suggested that the process of synaptic plasticity – a mechanism which underpins learning and memory encoding – is temporally dependent. In other words, the timing, or synchrony, of neural signalling is important – something eloquently realised in the phrase “cells which fire together, wire together”. For example, the soundtrack and imagery of a video will simultaneously activate you auditory and visual cortices respectively. But how does your brain “glue” together these two sensations within a memory? One hypothesis is that this would require a large scale synchronization of activity between the different sensory regions that can essentially serve to associate the activity of these regions into a common framework, something that may perhaps be visible in the EEG.
A focus on timing
A recent study in the journal Current Biology by Andrew Clouter, Kimron Shapiro and Simon Hanslmayr from the University of Birmingham in the UK has brought us a step closer to understanding this specific aspect of associative memory formation and drawing some relationship to EEG. In it they explored how the synchrony between information presented between audito
ry and visual information could influence the strength of the memories formed, and reflect in the entrainment of the EEG. To begin with 9 study participants with a movie clip and a soundtrack in which they manipulated the luminance of the movie image and the amplitude of the soundtrack in a manner (i.e. creating a flickering effect) that was either in phase or out of phase. Participants were then asked to recall which soundtrack went together with which movie clip. When the manipulation of the image luminance and sound amplitude were in phase (concomitant rather than time shifted), the accuracy of the memory tended to be better by about 20% suggesting that in general people were better at remembering when the modalities were somehow synced – timing matters.
A theta preference
More 
interesting was that this 20% difference in accuracy was only present when the flickering of the luminance and sound was presented at 4 Hz but not 1.7 Hz or 10.5 Hz. 4 Hz is the theta range and oscillations in the hippocampus at this frequency (measured in mice using more invasive high resolution techniques) have been previously associated with memory formation. Thus it appears that timing and presentation at certain frequencies matters but not others.
It is also worth noting though that given the large error bars in the graphs, some people obviously still performed pretty well on the memory test even when the flickering was at another frequency and the phase difference was high. It might therefore be of interest to see if the memory processes of different people are inclined towards different timing profiles.
See related post Intra person variability in the EEG
Stimulus timing in the EEG
Along with the presentation of the movies, they also recorded EEG activity in the auditory and visual regions of the cortex. Indeed they were able to see evidence that the EEG produced activity in those regions that mirrored the oscillations of the luminance in the visual cortex and the sound amplitude in the auditory cortex. When the flickering of sound and image were presented in phase, the EEG below 15 Hz was well matched across regions, when it was out of phase, the phase difference between the two regions had obvious relation to the phase shift between the two stimulus modalities. Thus timing of stimulus activity presents itself in the EEG in a measurable way.
Among other things, this emphasizes that simply studying the changes of a particular frequency band of the EEG in a single region is grossly insufficient to reveal the intricacies of complex phenomenon. Rather views into the broader network activity can deliver a more in depth understanding of the physiological correlates of cognitive function.
Bringing it back to memory
Clearly timing of stimuli from different modalities are reflected in the EEG, and timing matters for memory formation, but how does it tie together? The paper stops short of answering this but provides grist for hypothesis and the authors may find answers in looking more closely at individual trials and separately for each individual. For example, there is a wide range of phase differences across trials presented in the figure above. Are perhaps the phase relationships more tightly coupled for accurate trials? Are the phase relationships more tightly coupled for those individuals who do a better job at remembering which movie clip went together with which soundtrack? What comes out of such exploration may be well worth the effort.
