Memory, Perception, and the Medial Temporal Lobe: A Synthesis of Opinions

The current controversy in the literature concerning the putative role of the medial temporal lobe in perception has attracted so much attention because it challenges a fundamental tenet of a well-established view of the organization of memory in the brain. The medial temporal lobe memory system (MTLMS) view states that the medial temporal lobe is specialized for declarative/relational memory function including both episodic and semantic memory with little or no substantial role in perception (Eichenbaum and Cohen, 2001xFrom Conditioning to Conscious Recollection. Eichenbaum, H. and Cohen, N.J. See all References, Squire and Zola-Morgan, 1991xSquire, L.R. and Zola-Morgan, S. Science. 1991; 253: 1380–1386Crossref | PubMedSee all References). As we have both reviewed in the preceding articles, evidence from both animal and human studies has been used to argue that the MTL and in particular the perirhinal cortex is critical for certain forms of perception, often referred to as the perceptual-mnemonic hypothesis of MTL function. Although opinions remain highly polarized on this topic, there are several key theoretical and technical points that both sides can agree upon. Below we summarize/synthesize these points and outline several possible directions for future research that may help resolve this controversy.A central theoretical issue at the core of this controversy is the distinction between memory and perception. The ability to generate testable predictions that can differentiate the MTLMS hypothesis from the perceptual-mnemonic hypothesis of MTL function depends strongly on agreement about the tasks and specific testing methods that are used to examine perceptual functions in both animals and humans. For example, tasks in which trial-unique stimuli are available for simultaneous comparison would seem to place maximal demands on perceptual processes and minimal demands on memory (Lee et al., 2005axLee, A.C., Buckley, M.J., Pegman, S.J., Spiers, H., Scahill, V.L., Gaffan, D., Bussey, T.J., Davies, R.R., Kapur, N., Hodges, J.R., and Graham, K.S. Hippocampus. 2005; 15: 782–797Crossref | PubMed | Scopus (157)See all References, Shrager et al., 2006xShrager, Y., Gold, J.J., Hopkins, R.O., and Squire, L.R. J. Neurosci. 2006; 26: 2235–2240Crossref | PubMed | Scopus (69)See all References). Visual discrimination tasks where morphed stimuli are compared to a target stimulus presented next to the stimuli being discriminated also seem promising as a behavioral tool to minimize memory demands (Lee et al., 2005bxLee, A.C., Bussey, T.J., Murray, E.A., Saksida, L.M., Epstein, R.A., Kapur, N., Hodges, J.R., and Graham, K.S. Neuropsychologia. 2005; 43: 1–11Crossref | PubMed | Scopus (177)See all References, Shrager et al., 2006xShrager, Y., Gold, J.J., Hopkins, R.O., and Squire, L.R. J. Neurosci. 2006; 26: 2235–2240Crossref | PubMed | Scopus (69)See all References). The preferential looking paradigm used by Bartko et al. (2007)xBartko, S.J., Winters, B.D., Cowell, R.A., Saksida, L.M., and Bussey, T.J. J. Neurosci. 2007; 27: 2548–2559Crossref | PubMed | Scopus (85)See all ReferencesBartko et al. (2007) also follows this general strategy and is also potentially promising, though the possibility that rats used a working memory-like strategy to perform this task with large 3D objects must be addressed. Although a “memory-proof” task of perception that does not engage MTL-dependent memory functions has yet to be developed for animals, it will be important to use key elements from previous paradigms together with other yet-to-be developed tasks/approaches to develop improved tests of perception that do not engage the MTL-dependent memory.Another fundamental issue germane to this debate is whether it is even possible to test perceptual performance in the absence of learning or memory requirements. This is a challenging issue for a number of reasons. First, it is difficult to know what behavioral strategies animals or humans are using to solve a given task. So, for example, even if an animal performs a simple discrimination well, this does not guarantee that it will not apply a different strategy to the same task with much more difficult to discriminate stimuli. Second, the resolution of this question depends on the precise terms that are used to define “memory” and “perception” as independent of one another, and there is currently little agreement on these terms. Thus, one constructive outcome of this debate would be to force both sides to be more specific about their definition of these terms when discussing their own results. Unfortunately, these definitions often degenerate into arguments about semantics.Some of the difficulties outlined above might be circumvented, to some extent, by seeking converging evidence from other experimental methodologies. One particularly promising approach is behavioral neurophysiology. This approach can be used to test the key idea that while the perirhinal cortex may not be the only area that deals with feature conjunctions, it is a region “that contains perhaps the most complex conjunctive representations in the ventral visual stream” (Bussey and Saksida, 2002xBussey, T.J. and Saksida, L.M. Eur. J. Neurosci. 2002; 15: 355–364Crossref | PubMed | Scopus (152)See all ReferencesBussey and Saksida, 2002). Importantly, this method allows for the analysis of neural coding properties in the absence of any cognitive demands beyond viewing of the visual stimulus. Thus, one testable hypothesis consistent with the perceptual-mnemonic view is that neural activity specific to a particular category of highly complex stimuli would be seen exclusively (or at least more prominently) in the perirhinal cortex compared to upstream visual areas (i.e., visual area TE). It would also be possible to test whether neural correlates of complex or feature-ambiguous stimuli that are observed in upstream visual areas (such as area TE) are dependent on perirhinal cortex, by recording from area TE while perirhinal cortex is reversibly inactivated. Data from this kind of experiment would allow us to better define the kinds of visual stimuli processed in these areas as well as start to explore how these adjacent areas (i.e., the perirhinal cortex and area TE) interact in the processing of complex visual stimuli, both key issues in this debate.Another main point that most would agree on is the idea that the anatomical localization of the amnesic patients used in studies of the perception/memory debate must be fully characterized, including possible damage to lateral temporal cortex, in order to fully compare findings across studies. There remains substantial disagreement in the literature as to the success with which this anatomical localization has been carried out across various patient groups (Lee et al., 2005bxLee, A.C., Bussey, T.J., Murray, E.A., Saksida, L.M., Epstein, R.A., Kapur, N., Hodges, J.R., and Graham, K.S. Neuropsychologia. 2005; 43: 1–11Crossref | PubMed | Scopus (177)See all References, Shrager et al., 2006xShrager, Y., Gold, J.J., Hopkins, R.O., and Squire, L.R. J. Neurosci. 2006; 26: 2235–2240Crossref | PubMed | Scopus (69)See all References). To address this issue, consensus must be reached on the most appropriate methods for documenting and measuring the extent of brain damage in humans with MTL lesions, so that patient populations available to different research groups can be compared in terms of the extent and selectivity of their lesions. Questions about the extent of lesions have hampered progress in human neuropsychology and, to some extent, in studies with animals as well, although for different reasons. An effort to develop consensus criteria for imaging and volumetric protocols in human patients with MTL damage would help to reduce the concerns about the interpretability of studies based on the extent of lesions. This is essential because, ultimately, neuropsychology addresses the question of whether a structure is necessary for a particular aspect of cognition, which is the key question in understanding how neural dysfunction gives rise to cognitive impairment in disease.The concept that the MTL is specialized for memory and not for perception has been a highly influential one in the field, but it must withstand the test of scrutiny from subjects with highly selective brain damage and from more specific tests of perception. At this point, the evidence for a role of the MTL in perception is provocative but is still subject to alternative interpretations. However, with the addition of more specific tasks of perception, hypothesis-based neurophysiology studies together with strict criteria for how the extent of lesions are described in human neuropsychological studies, we feel that this controversy has a good chance of ultimately being resolved.