Memory has intrigued us for millenia, and is today one of the most active areas of neuroscience research. Much of this research has aimed to understand how memories are laid down, and a picture of how this happens is beginning to emerge. Hundreds of studies published over the past few decades provide evidence that memory formation involves widespread reorganization of connections in the brain.
The vast majority of memory research has therefore pertained to the neural processes which occur after the event that is being memorized. More recently, though, a number of studies have suggested that brain activity at the time of an event is experienced may play an important role in how memories of the event are encoded, and it is now thought that this activity determines the quality and strength of the memories.
A team of researchers from Germany now show that the activity which immediately precedes an event is also important for memory formation. In Proceedings of the National Academy of Sciences, they report the identification of a signature brain state which occurs just before the appearance of a visual stimulus and predicts the successful encoding of it. The findings point to ways in which the process of memory formation could be enhanced.
Sebastian Guderian and his colleagues recruited 24 healthy young participants and asked them to perform a simple test in which they made judgements about each item in a series of word lists. In one condition, they were told to press one key if a word contained two syllables, and another if it contained any other number of syllables. In the second condition, the keys were used to indicate whether a word was pleasant or unpleasant. After each test, they were given four moderately difficult mental arithmetic tasks, and then prompted to begin a free-recall test, in which they try to remember as many words as possible from the list they had just seen. As these tasks were performed, the researchers recorded the electrical activity of the participants’ brains using magnetoencephalography (MEG).
As expected, the participants were better able to recall the words which they had to judge as being either pleasant or unpleasant. These were studied in more detail, and processed according to their meaning; the words in the other lists were processed at a shallower level, because the participants were only required to indicate how many syllables they contained. But for both lists, the position of a word did not affect how successfully it was later recalled. Words at the top of the lists could not be selectively rehearsed, because they were quickly followed by the rest of the words in the list. And the arithmetic was included as a distractor task, to minimize the recency effect, whereby items at the end are retained in working in memory, and therefore better recalled.
Reconstructions of the current density of prestimulus theta waves, with the brain tilted to expose the ventromedial surface of the right medial temporal lobe (MTL). Words that were better remembered later on were associated with stronger theta currents in the MTL (left) than words that were forgotten (right). (From Guderain et al, 2009)
When the researchers analyzed the MEG recordings, they found that word recall was associated with particular patterns of brain waves. Words that were not remembered during free recall were associated with an increase in the amplitude of theta waves – “slow” oscillatory activity with a frequency of 3-7 Hertz – recorded from sensors lying over the occipital and temporal lobes, between 550-800 milliseconds after word presentation. By contrast, words that were successfully remembered were associated with an increase in theta wave amplitude in the medial temporal lobe, beginning 250 ms before word presentation.
The MEG data were then sorted into 5 separate groups of increasing theta wave amplitude. This showed that the rate of recall increased as a function of increasing prestimulus theta wave amplitude; in other words, the bigger the theta waves prior to presentation of a word, the better the subsequent recall of the word. This effect was observed in all 24 participants, and was found to be independent of the type of memory – it also occurred with the words that were judged according to how many syllables they contain, even though the overall recall rate for these words was lower. Thus, the rate of recall of a word can be predicted from the amplitude of theta waves in the medial temporal lobe just before presentation of that word.
One possible explanation for this effect is that theta wave amplitude in the hippocampus fluctuates randomly, so that words presented at a time which coincides with high amplitude waves were encoded properly and successfully recalled later on. But statistical tests showed that this was not the case. Another is that recall rates were related to levels of arousal, but this too was ruled out, because there was no significant difference in the reaction times across trials involving remembered and forgotten words. The increased theta amplitude measured prior to remembered words therefore seems to be generated at least partly by active processes, such as anticipation of a stimulus, or the intention to commit it to memory. This could be investigated in further studies in which the stimuli to be remembered are presented at unpredictable times.
How might an increase in the amplitude of theta waves enhance memory encoding? This pattern of electrical activity may reflect a state in which the hippocampus is especially conducive to synaptic plasticity the process by which connections between neurons are strengthened, and which is believed to be the cellular basis of learning and memory. This is supported by earlier experiments which show that rabbits learn better to associate two stimuli when those stimuli are presented at times of high theta activity in the hippocampus. The authors of the new study suggest that people could enhance their memory by using neurofeedback to optimize their hippocampal theta state.
Guderian, S., et al (2009). Medial temporal theta state before an event predicts episodic encoding success in humans Proc. Nat. Acad. Sci. DOI: 10.1073/pnas.0900289106.