Two kinds of memory signals in neurons of the human hippocampus

Significance Episodic memories represent the “what,” “when,” and “where” of specific episodes. In epilepsy patients, we detected single-unit activity reflecting episodic memory only in the hippocampus. This neural signal is sparsely coded and pattern-separated, consistent with predictions from neurocomputational models. We also detected single-unit activity reflecting a generic memory signal, coding whether an item is old or new without item-specific episodic information. Similar to concept cells, this generic repetition/novelty neural signal was found in multiple brain regions, including the hippocampus. In contrast, the item-specific signal was found only in the hippocampus. Our results indicate the coexistence of two memory signals in the human brain and suggest that the sparsely coded, hippocampus-specific signal is fundamental, whereas the often-studied generic signal is derivative.

neurons began ∼300 ms after stimulus onset and because nearly all behavioral responses occurred after 1 s.

Electrodes and Microwires Decisions. In the protocol used at the Barrow Neurological
Institute, decisions about whether to implant the depth electrodes and where to implant them are based solely on clinical criteria and are completely independent of decisions pertaining to microwire placement and recording.
Patients were selected for depth electrode implantation at the Barrow Neurological Institute following the protocol approved by the St. Joseph's Hospital Institutional Review Board. Patients are selected based on well-defined clinical criteria in line with generally recognized diagnostic practices (6,7). The decision to offer intracranial monitoring of seizures to a patient and the planning of the clinical study is made by the treating epileptologist with advice from the participants in the weekly case conference. Only those patients for whom noninvasive tests fail to narrow down the seizure focus sufficiently for adequate planning of the surgical resection have depth electrodes. Patients are told of the decision by their treating neurologists. They are then seen by their neurosurgeon. At this time, they are provided with a copy of the surgical consent forms and made aware of the risks and benefits in undertaking depth electrode implantation and intracranial monitoring as well as the subsequent surgical resection.
Thus, decisions about placement of the depth electrodes involve many physicians other than the potential clinical co-authors on these papers, Drs. Smith (the neurosurgeon) and Treiman, and are independent of any research concerns. No patient was ever referred for depth electrode placement by Dr. Treiman without the consensus of the EMU conference participants.
The placement of the depth electrodes in these four brain areas, including the anterior cingulate cortex and ventromedial prefrontal cortex, was a longstanding practice adopted by the clinical team in ~1995. These sites were implanted to check whether the patient's seizures had an extratemporal origin which would then change recommendations for further surgery. The The following is a photograph showing the scale of the clinically required depth electrode on the left with the microwires protruding to the right. The depth electrode is 1.8 mm in diameter and is typically inserted to a depth of 8-10 cm laterally though the temporal cortex. There are 9 microwires in the bundle, each of which is 38 µm in diameter and they protrude 5 mm from the tip of the depth electrode. The microwires thus represent little additional penetration of the brain tissue beyond what is clinically required.
Patients were provided a detailed written description of the nature of the microwires and the purpose of recording after meeting with the neurosurgeon to discuss the need for the depth electrode recordings and the surgery to implant the depth electrodes. They were asked to consider whether they would like to participate in the time between this meeting and their next meeting prior to surgery.
At that next meeting a member of the research team, typically Dr. Steinmetz, discussed the protocol with the patient and their family and addressed any questions they might have had about the microelectrodes, their placement, and the type of experimental tasks they would be asked to perform while in the epilepsy monitoring unit. At this meeting, which usually ranged between 15 to 30 minutes, the patients were explicitly asked if they would like to see an example of the electrodes and microwires. In most cases, the patients did wish to see and touch them.
Only after all concerns were discussed were the patients asked if they would agree to microwire placement. If they did agree, they then signed the written statement of consent to microwire placement and recording.
In a recent article noting that it now seems appropriate to use microwire recordings to clinically benefit the patients, Chari, Thornton, Tisdall and Scott (8)  In all, we always ensured that the placement of microwires and subsequent recording presented little additional risk to them. We also ensured that the patients fully understood the nature of the recordings and microwires and were completely free to make their own decision on whether or not to participate based upon their desire to participate in such scientific research.
Filtering and Event Detection. Extracellular potentials were recorded from the tips of the microwires using techniques previously described (9) and digitized at 29,412 Hz with 16-bit resolution. Possible action potential events (APs) were detected using digital filtering and thresholding (10). Because more than one neuron may be recorded near any given electrode, APs were sorted into several clusters of similar waveform shape using the open-source clustering program KlustaKwik (Klustakwik.sf.net). After sorting, each cluster was graded as being noise, multiunit activity (MUA), or single-unit activity (SUA) based on criteria such as the waveform shape, size of the waveform relative to noise, evidence of a refractory interval, and lack of powerline interference, using the criteria described previously (9).
In our experience, this technique produces results comparable to prior reports in other laboratories (10) in terms of recorded waveform shapes, interspike intervals, and firing rates.
While it is important to note that these and other reports of human single-unit recordings (11) do not achieve the quality of unit separation achievable in animal recordings (12) they nonetheless represent neural activity at a much finer spatial and temporal scale than is achievable using other methods such as fMRI.
6 Behavioral Performance at different lags. Each of the 6 lags (i.e., with 0, 1, 3, 7, 15, or 31 intervening words) had an equal number of trials for both the visual and auditory sessions. For the visual sessions, 120 unique words were repeated, and 20 words were repeated at each lag.
For the auditory sessions, 300 unique words were repeated, and 50 words were repeated at each lag. As is typical, recognition performance was better at shorter lags. At the six lags, patients were 87.5%, 81.6%, 81.0%, 75.9%, 72.5%, and 70.3% correct in recognizing a previously presented word, respectively. There was no ceiling effect at 0 or 1 lag. This is probably because it was a continuous recognition task in which, for each trial, participants needed to hold as many previous words as possible in their working memory or else they relied on long-term memory.
3. Two raster plots of raw spikes for the item-specific memory signal.