Plugging in to Human Memory: Advantages, Challenges, and Insights from Human Single-Neuron Recordings
2019; Cell Press; Volume: 179; Issue: 5 Linguagem: Inglês
10.1016/j.cell.2019.10.016
ISSN1097-4172
Autores Tópico(s)Neural dynamics and brain function
ResumoWe describe single-neuron recordings in the human hippocampal formation, performed in epileptic patients for clinical reasons, and highlight their advantages, challenges, and limitations compared with non-invasive recordings in humans and invasive recordings in animals. We propose a unified framework to explain different findings—responses to novel stimuli, spatial locations, and specific concepts—linking the rodent and human literature regarding the function of the hippocampal formation. Moreover, we propose a model of how memories are encoded in this area, suggesting that the context-independent, invariant coding by concept cells may provide a uniquely human neural mechanism underlying memory representations. We describe single-neuron recordings in the human hippocampal formation, performed in epileptic patients for clinical reasons, and highlight their advantages, challenges, and limitations compared with non-invasive recordings in humans and invasive recordings in animals. We propose a unified framework to explain different findings—responses to novel stimuli, spatial locations, and specific concepts—linking the rodent and human literature regarding the function of the hippocampal formation. Moreover, we propose a model of how memories are encoded in this area, suggesting that the context-independent, invariant coding by concept cells may provide a uniquely human neural mechanism underlying memory representations. In the 1980s, zoologist Hans Kummer reported a now very famous observation of a female Hamadrya baboon grooming with a young male hiding behind a big rock but keeping part of the body visible to her own male, who was feeding several meters away, unaware of the situation (Whiten and Byrne, 1988Whiten A. Byrne R.W. Tactical deception in primates.Behav. Brain Sci. 1988; 11: 233-273Crossref Google Scholar). This and similar types of behaviors have been offered as evidence that non-human primates have a "theory of mind"; that is, the ability to understand other subjects' thoughts. However, this interpretation has been disputed by behavioralists (see comments in Whiten and Byrne, 1988Whiten A. Byrne R.W. Tactical deception in primates.Behav. Brain Sci. 1988; 11: 233-273Crossref Google Scholar, who argue that the animal may have learned to act this way without truly understanding why; that means, without necessarily wondering what her male was thinking. The problem is that we cannot get into the animal's head or simply ask her why she hid behind the rock. In fact, pinning down the real motives or thoughts of animals is difficult and requires well-controlled paradigms (one such study showed decades later that non-human primates do indeed have a theory of mind [Krupenye et al., 2016Krupenye C. Kano F. Hirata S. Call J. Tomasello M. Great apes anticipate that other individuals will act according to false beliefs.Science. 2016; 354: 110-114Crossref PubMed Google Scholar]). The same applies to memory, particularly to episodic memory (i.e., the memory of our experiences) because we cannot interrogate animals about their thoughts and recollections. A key challenge in neuroscience is to understand how the firing of neurons underlies behavior. However, advances in this area face some very basic limitations. On one hand, non-invasive recording techniques—e.g., electroencephalography (EEG), magnetoencephalography (MEG), and fMRI—are used with human subjects for obvious ethical reasons, but, although these methods have provided insights into the activation of brain areas during different tasks, they can only offer an indirect and vague measure of the activity of individual neurons (Logothetis, 2008Logothetis N.K. What we can do and what we cannot do with fMRI.Nature. 2008; 453: 869-878Crossref PubMed Scopus (1737) Google Scholar). On the other hand, invasive recordings provide direct access to study the firing of multiple neurons but can usually only be performed in animals, and, as the story of Kummer illustrates, the lack of direct verbal feedback limits our understanding of what is going on in the animal's brain. Moreover, the types of experiments and questions that can be addressed with animals are limited because they need extensive reward-driven training to perform different tasks, far from the natural conditions of how these behaviors occur in real-life situations. In very particular cases, however, it is possible to perform invasive recordings in human subjects for clinical reasons. This is the case with patients suffering from epilepsy refractory to medication, who are implanted with intracranial electrodes to determine the seizure-originating area and evaluate the possibility of its surgical resection (Rey et al., 2015aRey H.G. Ison M.J. Pedreira C. Valentin A. Alarcon G. Selway R. Richardson M.P. Quian Quiroga R. Single-cell recordings in the human medial temporal lobe.J. Anat. 2015; 227: 394-408Crossref Scopus (20) Google Scholar), offering the extraordinary opportunity to record the activity of multiple single neurons in awake and behaving human subjects performing different tasks. (Single-cell recordings are also performed during deep brain stimulation [DBS], and we refer to Engel et al., 2005Engel A.K. Moll C.K.E. Fried I. Ojemann G.A. Invasive recordings from the human brain: clinical insights and beyond.Nat. Rev. Neurosci. 2005; 6: 35-47Crossref PubMed Scopus (222) Google Scholar for a review of these studies, which will not be covered here.) The first recordings of individual neurons in the human brain were performed in the 1950s, using a glass pipette attached to a micromanipulator during epilepsy surgery (Ward and Thomas, 1955Ward A.A. Thomas L.B. The electrical activity of single units in the cerebral cortex of man.Electroencephalogr. Clin. Neurophysiol. 1955; 7: 135-136Abstract Full Text PDF PubMed Google Scholar). Later on, in the 1970s, a procedure was developed to record from multiple microwires that were inserted through hollow-depth intracranial electrodes protruding a few millimeters from their end (Babb et al., 1973Babb T.L. Carr E. Crandall P.H. Analysis of extracellular firing patterns of deep temporal lobe structures in man.Electroencephalogr. Clin. Neurophysiol. 1973; 34: 247-257Abstract Full Text PDF PubMed Scopus (0) Google Scholar), a design that it is still used today (Figures 1A–1C). Contacts placed along the depth electrodes allow recording of intracranial electroencephalographic (iEEG) data used for clinical assessment of the patients, whereas the microwires provide recordings of multiple single neurons and local field potentials (LFPs) (Figures 1D and 1E). Recording sites often cover the medial temporal lobe (MTL; the hippocampal formation and its surrounding cortex) because of the involvement of this area in different forms of epilepsy (Niedermeyer, 1993Niedermeyer E. Epileptic seizure disorders.in: Lopes da Silva F. Niedermeyer E. Electroencephalography. Williams and Wilkins, 1993: 461-564Google Scholar). Subjects remain with the electrodes implanted for about a week and are monitored to record a sufficient number of spontaneous seizures to evaluate an eventual surgical resection of the epileptic focus. Compared with non-invasive studies, the key advantage of single-neuron recordings is the possibility of having access to the activity of individual neurons, which can be measured only indirectly with non-invasive methods. Let us illustrate this with two concrete cases. First, it is common that MTL neurons respond sparsely to very few pictures (Quian Quiroga et al., 2007Quian Quiroga R. Reddy L. Koch C. Fried I. Decoding visual inputs from multiple neurons in the human temporal lobe.J. Neurophysiol. 2007; 98: 1997-2007Crossref PubMed Scopus (0) Google Scholar). Because of a general lack of topographic organization in the MTL (i.e., responses are not spatially clustered, and nearby neurons fire to completely different stimuli; De Falco et al., 2016De Falco E. Ison M.J. Fried I. Quian Quiroga R. Long-term coding of personal and universal associations underlying the memory web in the human brain.Nat. Commun. 2016; 7: 13408Crossref PubMed Scopus (13) Google Scholar), there is not a common and localized activation that can be observed at the more macroscopic level of fMRI or EEG/MEG recordings, and, therefore, these responses are only identified at the single-neuron level. Second, besides providing information about neuronal responses that cannot be seen with non-invasive methods, single-neuron recordings can also validate and provide further mechanistic evidence of fMRI and EEG/MEG findings. For example, fMRI studies have consistently shown the presence of preferential responses to scenes in the parahippocampal place area (PPA) (Epstein and Kanwisher, 1998Epstein R. Kanwisher N. A cortical representation of the local visual environment.Nature. 1998; 392: 598-601Crossref PubMed Scopus (1820) Google Scholar). However, fMRI recordings cannot distinguish between different mechanisms that can produce these responses: (1) each PPA neuron may respond sparsely to one or relatively few scenes, giving category scene responses when averaging the activity of neighboring neurons constituting each voxel; (2) PPA neurons may be tuned to visual features that are more prevalent in scene images, showing scene responses in the population average but some visual feature tuning, rather than scene selectivity, at the single-neuron level; (3) PPA neurons may be scene-selective, responding preferentially to pictures of scenes. Analysis of about 2,000 human MTL neurons clearly showed that the latter was the case; parahippocampal neurons had a tendency to respond to scenes (Mormann et al., 2017Mormann F. Kornblith S. Cerf M. Ison M.J. Kraskov A. Tran M. Knieling S. Quian Quiroga R. Koch C. Fried I. Scene-selective coding by single neurons in the human parahippocampal cortex.Proc. Natl. Acad. Sci. USA. 2017; 114: 1153-1158Crossref PubMed Scopus (0) Google Scholar; Figure 2A) with much broader category tuning compared with the selective responses found for known people (Quian Quiroga et al., 2007Quian Quiroga R. Reddy L. Koch C. Fried I. Decoding visual inputs from multiple neurons in the human temporal lobe.J. Neurophysiol. 2007; 98: 1997-2007Crossref PubMed Scopus (0) Google Scholar). Compared with recordings with animals, a first obvious advantage is that, if the ultimate goal is to understand the human brain (although this need not necessarily be the case), then, by performing recordings directly in humans, we can avoid the potentially false assumption of similar brain functioning in animal models and humans. Another advantage of human single-neuron recordings is the possibility of communicating with and getting direct feedback from the subjects, which allows us to perform experiments that cannot be done in other animals. This is particularly the case when studying internally generated top-down activations, such as those arising during memory recall (Gelbard-Sagiv et al., 2008Gelbard-Sagiv H. Mukamel R. Harel M. Malach R. Fried I. Internally generated reactivation of single neurons in human hippocampus during free recall.Science. 2008; 322: 96-101Crossref PubMed Scopus (270) Google Scholar, Ison et al., 2015Ison M.J. Quian Quiroga R. Fried I. Rapid encoding of new memories by individual neurons in the human brain.Neuron. 2015; 87: 220-230Abstract Full Text Full Text PDF PubMed Scopus (0) Google Scholar), imagery (Kreiman et al., 2000Kreiman G. Koch C. Fried I. Imagery neurons in the human brain.Nature. 2000; 408: 357-361Crossref PubMed Scopus (0) Google Scholar), or voluntary control of the neuron's firing based on the subject's thoughts (Cerf et al., 2010Cerf M. Thiruvengadam N. Mormann F. Kraskov A. Quian Quiroga R. Koch C. Fried I. On-line, voluntary control of human temporal lobe neurons.Nature. 2010; 467: 1104-1108Crossref PubMed Scopus (91) Google Scholar). Another key advantage is the possibility of directly explaining the experiments to the subjects without the need of extensive reward-driven training and the ensuing potential caveats of overtraining effects that could influence the neurons' responses after months of performing the same task. Moreover, communication with the subjects permits tuning the experiments according to their background and interests. For example, Figure 2B shows the responses of a neuron in a subject interested in mathematics that fired to different equations and math-related stimuli, whereas Figure 2C shows a neuron's responses to "Mr. T," a character in the film Rocky III, in a subject who was a fan of this movie. The rationale for presenting equations (among other things) to the first subject and characters from the Rocky films to the second was that we expected to find more responses to personally relevant things, as was shown to be the case from analysis of a large number of responses (Viskontas et al., 2009Viskontas I.V. Quian Quiroga R. Fried I. Human medial temporal lobe neurons respond preferentially to personally relevant images.Proc. Natl. Acad. Sci. USA. 2009; 106: 21329-21334Crossref PubMed Scopus (0) Google Scholar). Human single-neuron recordings offer clear advantages but have also several limitations, mainly because of recording constraints and the fact that these experiments are performed with patients in a clinical environment. A major limitation is the availability of patients. Most hospitals performing these recordings have relatively few implantations a year, and it may take several years to get a sufficient number of neurons to have statistically sound results. Moreover, the time to perform experiments with each patient is also limited, and it is not always possible to consider all of the control experiments that one would like to do. In addition, recordings are performed in a clinical environment, which can be very noisy, and there is relatively little time to sort out technical issues compared with a standard laboratory environment. The fact that recordings are done in patients with epilepsy also raises the concern that the obtained results may reflect different aspects of this pathology rather than normal brain functioning. This is, however, very unlikely given that similar types of responses have been obtained in recordings close to the seizure-originating area and in more distant areas, including the non-seizure-originating hemisphere (Mormann et al., 2008Mormann F. Kornblith S. Quian Quiroga R. Kraskov A. Cerf M. Fried I. Koch C. Latency and selectivity of single neurons indicate hierarchical processing in the human medial temporal lobe.J. Neurosci. 2008; 28: 8865-8872Crossref PubMed Scopus (123) Google Scholar). Moreover, results are similar for patients with different types of epilepsy involving different pathophysiological mechanisms (Niedermeyer, 1993Niedermeyer E. Epileptic seizure disorders.in: Lopes da Silva F. Niedermeyer E. Electroencephalography. Williams and Wilkins, 1993: 461-564Google Scholar). In addition, epileptic activity is, in principle, expected to produce an increase in neural excitability and connectivity, which would give a global increase in the neurons' responsiveness, contrary to the very high selectivity observed in these recordings (Quian Quiroga et al., 2007Quian Quiroga R. Reddy L. Koch C. Fried I. Decoding visual inputs from multiple neurons in the human temporal lobe.J. Neurophysiol. 2007; 98: 1997-2007Crossref PubMed Scopus (0) Google Scholar). Results could also be attributed to effects of the medication taken by the patients. This is particularly a concern when considering, for example, the relatively late onset of MTL neuron responses compared with response onsets in animals (Mormann et al., 2008Mormann F. Kornblith S. Quian Quiroga R. Kraskov A. Cerf M. Fried I. Koch C. Latency and selectivity of single neurons indicate hierarchical processing in the human medial temporal lobe.J. Neurosci. 2008; 28: 8865-8872Crossref PubMed Scopus (123) Google Scholar). However, different patients have different medications and dosages, and, furthermore, medication is gradually tapered down during the time the patient is in the hospital to increase the chances of recording seizures. Because similar results are obtained in different patients and at different days of the intervention, the effects of medication in the MTL responses can be ruled out. A caveat of human single neuron recordings is their limited coverage compared with non-invasive techniques. The location of the intracranial electrodes is always determined by clinical criteria. Consequently, scientist do not have—and should not have—a say in decisions about the implantation of the electrodes, which may not necessarily cover the key areas involved in the processes under study. Moreover, as with chronic recordings, electrodes cannot be externally moved to search for responsive neurons, and millimeter variations in the electrode implantation can mean the difference between obtaining and not obtaining single-neuron recordings. However, because the electrodes are fixed, there are no potential biases as there can be with acute recordings: moving the electrodes and targeting easily identifiable neurons with high firing rates can lead to sparsely firing neurons being overlooked (Shoham et al., 2006Shoham S. O'Connor D.H. Segev R. How silent is the brain: is there a "dark matter" problem in neuroscience?.J. Comp. Physiol. A Neuroethol. Sens. Neural Behav. Physiol. 2006; 192: 777-784Crossref PubMed Scopus (116) Google Scholar). Recording sites typically include the MTL because of the involvement of this area in different forms of epilepsy (Niedermeyer, 1993Niedermeyer E. Epileptic seizure disorders.in: Lopes da Silva F. Niedermeyer E. Electroencephalography. Williams and Wilkins, 1993: 461-564Google Scholar). This is ideal to study memory processes, given the well-documented role of the MTL in declarative memory (Squire and Zola-Morgan, 1991Squire L.R. Zola-Morgan S. The medial temporal lobe memory system.Science. 1991; 253: 1380-1386Crossref PubMed Google Scholar). However, the study of MTL neurons provides only a limited picture of memory functions, which should ideally also consider interactions with the diencephalon (Aggleton and Brown, 1999Aggleton J.P. Brown M.W. Episodic memory, amnesia, and the hippocampal-anterior thalamic axis.Behav. Brain Sci. 1999; 22 (discussion 444–489): 425-444Crossref PubMed Scopus (1335) Google Scholar) and neocortical areas (Eichenbaum, 2017Eichenbaum H. Prefrontal-hippocampal interactions in episodic memory.Nat. Rev. Neurosci. 2017; 18: 547-558Crossref PubMed Scopus (83) Google Scholar, Fletcher and Henson, 2001Fletcher P.C. Henson R.N. Frontal lobes and human memory: insights from functional neuroimaging.Brain. 2001; 124: 849-881Crossref PubMed Google Scholar, Sekeres et al., 2018Sekeres M.J. Winocur G. Moscovitch M. The hippocampus and related neocortical structures in memory transformation.Neurosci. Lett. 2018; 680: 39-53Crossref PubMed Scopus (13) Google Scholar). This is particularly important to study memory consolidation and the interplay of these areas in the coding of episodic and semantic memories (Moscovitch et al., 2005Moscovitch M. Rosenbaum R.S. Gilboa A. Addis D.R. Westmacott R. Grady C. McAndrews M.P. Levine B. Black S. Winocur G. Nadel L. Functional neuroanatomy of remote episodic, semantic and spatial memory: a unified account based on multiple trace theory.J. Anat. 2005; 207: 35-66Crossref PubMed Scopus (554) Google Scholar, Squire and Zola-Morgan, 1991Squire L.R. Zola-Morgan S. The medial temporal lobe memory system.Science. 1991; 253: 1380-1386Crossref PubMed Google Scholar). It is, however, possible to extract some information about neocortical activations from the iEEG contacts of the depth electrodes (Figure 1). In this respect, of particular interest is analysis of high-frequency oscillations, which correlate with local neuronal activity (Fisch et al., 2009Fisch L. Privman E. Ramot M. Harel M. Nir Y. Kipervasser S. Andelman F. Neufeld M.Y. Kramer U. Fried I. Malach R. Neural "ignition": enhanced activation linked to perceptual awareness in human ventral stream visual cortex.Neuron. 2009; 64: 562-574Abstract Full Text Full Text PDF PubMed Scopus (134) Google Scholar, Lachaux et al., 2012Lachaux J.P. Axmacher N. Mormann F. Halgren E. Crone N.E. High-frequency neural activity and human cognition: past, present and possible future of intracranial EEG research.Prog. Neurobiol. 2012; 98: 279-301Crossref PubMed Scopus (177) Google Scholar, Rich and Wallis, 2017Rich E.L. Wallis J.D. Spatiotemporal dynamics of information encoding revealed in orbitofrontal high-gamma.Nat. Commun. 2017; 8: 1139Crossref PubMed Scopus (1) Google Scholar, Watson et al., 2018Watson B.O. Ding M. Buzsáki G. Temporal coupling of field potentials and action potentials in the neocortex.Eur. J. Neurosci. 2018; 48: 2482-2497Crossref Scopus (19) Google Scholar) although, of course, not giving single-neuron resolution, as with implanted microwires (e.g., to estimate stimulus selectivity; Rey et al., 2014Rey H.G. Fried I. Quian Quiroga R. Timing of single-neuron and local field potential responses in the human medial temporal lobe.Curr. Biol. 2014; 24: 299-304Abstract Full Text Full Text PDF PubMed Scopus (28) Google Scholar). Another caveat is that it is not possible to give a precise location of the microwires used for human recordings. Animal studies, particularly with rodents, have offered clear evidence of distinct roles of substructures within the hippocampus (see, e.g., Treves and Rolls, 1994Treves A. Rolls E.T. Computational analysis of the role of the hippocampus in memory.Hippocampus. 1994; 4: 374-391Crossref PubMed Scopus (722) Google Scholar). But with human single-cell recordings, it is difficult to delineate the hippocampal substructures in the co-registered MRI scans (Wisse et al., 2017Wisse L.E.M. Daugherty A.M. Olsen R.K. Berron D. Carr V.A. Stark C.E.L. Amaral R.S.C. Amunts K. Augustinack J.C. Bender A.R. et al.Hippocampal Subfields GroupA harmonized segmentation protocol for hippocampal and parahippocampal subregions: Why do we need one and what are the key goals?.Hippocampus. 2017; 27: 3-11Crossref PubMed Scopus (32) Google Scholar), and it is also very difficult to visualize the microwires in the CT scans (Figures 1B and 1C) (and even if we could, we would not know which microwire is which). Another major limitation is given by the number of microwires implanted for single-neuron recordings. MTL neurons tend to show very sparse responses, firing to relatively few stimuli (Mormann et al., 2008Mormann F. Kornblith S. Quian Quiroga R. Kraskov A. Cerf M. Fried I. Koch C. Latency and selectivity of single neurons indicate hierarchical processing in the human medial temporal lobe.J. Neurosci. 2008; 28: 8865-8872Crossref PubMed Scopus (123) Google Scholar, Quian Quiroga et al., 2007Quian Quiroga R. Reddy L. Koch C. Fried I. Decoding visual inputs from multiple neurons in the human temporal lobe.J. Neurophysiol. 2007; 98: 1997-2007Crossref PubMed Scopus (0) Google Scholar). Consequently, it is difficult to trigger the neurons' responses (that is why we use the "screening sessions" described below), and it is very unlikely to record simultaneously from two or more neurons responding to a particular stimulus, although it is still possible to infer properties at the population level by using statistical arguments (e.g., Waydo et al., 2006Waydo S. Kraskov A. Quian Quiroga R. Fried I. Koch C. Sparse representation in the human medial temporal lobe.J. Neurosci. 2006; 26: 10232-10234Crossref PubMed Scopus (97) Google Scholar). Moreover, the number of identified units can be increased using optimal spike sorting algorithms (see next section) (Rey et al., 2015bRey H.G. Pedreira C. Quian Quiroga R. Past, present and future of spike sorting techniques.Brain Res. Bull. 2015; 119: 106-117Crossref PubMed Scopus (112) Google Scholar) and, particularly, new electrode designs. Spectacular advances have been made in the design of electrodes used for animal studies (e.g., Jun et al., 2017Jun J.J. Steinmetz N.A. Siegle J.H. Denman D.J. Bauza M. Barbarits B. Lee A.K. Anastassiou C.A. Andrei A. Aydın Ç. et al.Fully integrated silicon probes for high-density recording of neural activity.Nature. 2017; 551: 232-236Crossref PubMed Scopus (77) Google Scholar), but we have essentially been using the same type of electrodes for human single-cell recordings since the 1970s (Babb et al., 1973Babb T.L. Carr E. Crandall P.H. Analysis of extracellular firing patterns of deep temporal lobe structures in man.Electroencephalogr. Clin. Neurophysiol. 1973; 34: 247-257Abstract Full Text PDF PubMed Scopus (0) Google Scholar), in spite of the fact that progress in this area is likely to have a large effect on the recording conditions. Using chronic recordings in animals, it has been shown that it is possible to record from the same neurons during several days (Dhawale et al., 2017Dhawale A.K. Poddar R. Wolff S.B. Normand V.A. Kopelowitz E. Ölveczky B.P. Automated long-term recording and analysis of neural activity in behaving animals.eLife. 2017; 6: e27702Crossref PubMed Scopus (6) Google Scholar, Okun et al., 2016Okun M. Lak A. Carandini M. Harris K.D. Long term recordings with immobile silicon probes in the mouse cortex.PLoS ONE. 2016; 11: e0151180Crossref Scopus (27) Google Scholar). Tracking neurons over days is indeed critical for human MTL recordings because it allows us to assess the stability/plasticity of the responses and study consolidation mechanisms. In particular, there has been a long ongoing dispute about the role of the MTL in memory coding. Supporters of the standard consolidation model (Squire et al., 2015Squire L.R. Genzel L. Wixted J.T. Morris R.G. Memory consolidation.Cold Spring Harb. Perspect. Biol. 2015; 7: a021766Crossref PubMed Scopus (93) Google Scholar, Squire and Zola-Morgan, 1991Squire L.R. Zola-Morgan S. The medial temporal lobe memory system.Science. 1991; 253: 1380-1386Crossref PubMed Google Scholar) argue that the MTL encodes memories only during learning and not after their consolidation in the neocortex, whereas supporters of the multiple trace theory (Nadel and Moscovitch, 1997Nadel L. Moscovitch M. Memory consolidation, retrograde amnesia and the hippocampal complex.Curr. Opin. Neurobiol. 1997; 7: 217-227Crossref PubMed Scopus (1141) Google Scholar, Sekeres et al., 2018Sekeres M.J. Winocur G. Moscovitch M. The hippocampus and related neocortical structures in memory transformation.Neurosci. Lett. 2018; 680: 39-53Crossref PubMed Scopus (13) Google Scholar) argue that the MTL continues to remain critical for (episodic) memory after learning, thus providing a long-term representation. Evidence backing one or the other theory comes mainly from two sources: behavioral studies in patients with lesions, which, because of the variability of the precise location and extent of the lesions, have provided mixed results (Moscovitch et al., 2005Moscovitch M. Rosenbaum R.S. Gilboa A. Addis D.R. Westmacott R. Grady C. McAndrews M.P. Levine B. Black S. Winocur G. Nadel L. Functional neuroanatomy of remote episodic, semantic and spatial memory: a unified account based on multiple trace theory.J. Anat. 2005; 207: 35-66Crossref PubMed Scopus (554) Google Scholar), and non-invasive (fMRI) recordings (Moscovitch et al., 2005Moscovitch M. Rosenbaum R.S. Gilboa A. Addis D.R. Westmacott R. Grady C. McAndrews M.P. Levine B. Black S. Winocur G. Nadel L. Functional neuroanatomy of remote episodic, semantic and spatial memory: a unified account based on multiple trace theory.J. Anat. 2005; 207: 35-66Crossref PubMed Scopus (554) Google Scholar), but this technique cannot directly assess, at the single-cell level, the stability and plasticity of neuronal representations. Tracking neurons in time is a challenging task with recordings in epileptic patients (and patients typically have the electrodes implanted for no longer than a week). This is because, in the clinical environment where the experiments are performed, the signal and noise conditions can change abruptly, and electrodes may also move; for example, when the patients have seizures with abrupt contractions. In spite of this limitation, first results show that it is at least possible to track neurons producing similar responses on consecutive days (Niediek et al., 2016Niediek J. Boström J. Elger C.E. Mormann F. Reliable analysis of single-unit recordings from the human brain under noisy conditions: tracking neurons over hours.PLoS ONE. 2016; 11: e0166598Crossref PubMed Scopus (3) Google Scholar). However, it is still difficult to quantitatively assess the stability and plasticity of the neural representations—i.e., what the neurons fire to—across days because different responses could arise from different neurons being recorded. To address this issue, it is important to perform continuous 24/7 recordings and track the neurons' properties (e.g., spike shape, firing characteristics) to check their identity, ensuring that any changes in the neurons' properties are relatively smooth (Harris et al., 2016Harris K.D. Quian Quiroga R. Freeman J. Smith S.L. Improving data quality in neuronal population recordings.Nat. Neurosci. 2016; 19: 1165-1174Crossref PubMed Google Scholar). In this respect, it should be noted that, although it is difficult to record from the same neurons over several days in humans, it is, however, possible to infer whether responses are created de novo during the task or whether they reflect a long-term representation. The latter seems to be the case, given that MTL responses are observed during passive viewing the first time the patient sees a picture of a particular person (or place, animal, etc.), meaning that the neuron was already encoding this person before the experiment took place (Pedreira et al., 2010Pedreira C. Mormann F. Kraskov A. Cerf M. Fried I. Koch C. Quian
Referência(s)