Research
The Human Memory Lab is interested in a wide range of research topics related to memory. Please see below for some of our broad research interests.
Amnesia and the Medial Temporal Lobe
Since the discovery of patients like HM, it has been known that the medial temporal lobes (MTL) are essential for normal episodic memory. Over the past 20 years, the Human Memory Lab has aimed to uncover how different regions in the MTL support the processes involved in memory; essentially looking to determine the brain’s wiring diagram that underlies our ability to remember past events. In the late 1990s and early 2000s, we discovered that patients with damage to the hippocampus exhibited selective deficits in recollection, whereas damage to the surrounding perirhinal cortex lead to deficits in familiarity [1,2,3]. These discoveries directly challenged classical models of MTL function that treated these regions as a single unified memory system, and it converged with a growing body of work from nonhuman primates and rodents, which ultimately led to the development of our current hierarchical models of medial temporal lobe function. In addition, we have discovered that some regions in the MTL are critical for memory processes traditionally thought to be supported by cortical regions outside of the MTL, such as conceptual implicit memory [4]. Our more recent work indicates that these MTL structures may also contribute to visual short-term memory and complex perception tasks that have traditionally been thought to be independent of the MTL [5].
Neuroimaging and Memory
The medial temporal lobe works in conjunction with a complex set of brain networks to support different memory and cognitive processes. Early work using fMRI indicated that recollection was supported not only by the hippocampus, but by a network of brain regions including the parahippocampal cortex and various other cortical and subcortical regions; reflecting what is now referred to as the canonical recollection network [1,2]. Moreover, with the development of higher spatial resolution imaging protocols [3] we were able to separate the contribution of various MTL subregions to different processes involved in recognition, which has played critical role in testing competing models of MTL function [3,4]. These results largely converged with the anatomical results from our patient lesion studies which showed the roles of the hippocampus and perirhinal cortex. In addition, these imaging results revealed the involvement of a much broader neural network, and pointed to the importance of regions that we had not been able to examine in lesion studies. Current studies are examining the functional roles that different subregions within the hippocampus such as DG and CA1 play in memory and cognition.
Models of the Medial Temporal Lobe
The findings from our amnesia and neuroimaging studies have challenged traditional models of the medial temporal lobe; these traditional models had treated this structure as reflecting a single unified memory system. This has led us and several other labs to develop a new generation of models [1-4]. To account for these results, we have incorporated the basic assumptions of the dual process signal detection model with neuroanatomical findings, and proposed what we called the Binding of Items and Contexts (BIC) model [1]. In addition, we have utilized neurocomputational approaches to incorporate the functions of hippocampal subfields into our models [3,4]. Although these models are not without their critics, they do account for a larger set of findings from across human and animal studies. In addition, these approaches are particularly powerful in the sense that they make important new predictions about not only memory, but about perception and visual-working memory.
The Dual Process Signal Detection Model of Memory
Recognition memory judgments can be based either on the recollection of qualitative information about a previous episode, or on assessments of stimulus familiarity. In 1994, Dr. Yonelinas argued for a simple quantitative dual process model that characterized familiarity as a signal detection process whereby studied items are temporally strengthened, but in addition, some limited proportion of studied items are also recollected [1,2]. The original model has evolved over time, and several elaborations and extensions of the model have been proposed by various different labs [3,4]. Current work aims to bridge the model with neurocomputational models of the MTL.
Aging and Alzheimer’s Disease
Memory declines are common in normal aging and in Alzheimer’s disease. Developing methods to differentiate between different age-related changes in memory, and for early detection of these memory changes, is critical for effective treatment. Normal aging is associated with selective reductions in recollection that leave familiarity largely unaffected, whereas early Alzheimer’s disease is associated with declines in both recollection and familiarity [1,2]. Moreover, age related reductions in recollection are related to decreases in hippocampal integrity, as well as changes in the integrity of frontal lobe white matter, and alterations in widespread functional connectivity [3,4]. Current work in the lab is building on new findings, which suggest that age related hippocampal deficits in recollection might be partially overcome by promoting unitization strategies that utilize cortically based associative memory [5].
Emotion and Stress
The most memorable events of our lives are often those that are emotionally arousing. We have shown that the emotional advantages in memory emerge slowly over time, and appear to preferentially impact some forms of memory over others (e.g., item recollection rather than familiarity or context recollection), and they appear to rely crucially on the amygdala, but not on the hippocampus [1]. Current work examines how emotion and acute stress impact the various processes supporting memory, and aims to incorporate those insights into emerging models of MTL function [2,3,4].
Psychophysical Methods
In order to make advances in any science, it is necessary to first develop measurement methods and analytic tools that allow one to accurately characterize the phenomenon of interest. For example, measurement tools based on signal detection theory have proven essential in separating processes involved in perception and memory from those related to decision criterion and response bias. Working with Dr. Larry Jacoby, we advanced a ‘process dissociation’ framework [1] that aimed to separate the automatic and consciously controlled influences of memory. This method has now been applied across numerous different cognitive domains and has had a major impact on the field [2]. In addition, the Human Memory Lab has developed a method of analyzing receiver operating characteristics (ROCs) in order to separate the contribution or recollection and familiarity-based recognition memory processes [3]. This approach is now widely applied in studies of memory in various patient populations, as well as in rodents and nonhuman primates – which is proving important in bridging animal and clinical research domains. Other psychophysical methods advanced in the lab include subjective report methods as well as second-choice methods [4]. Current work aims to extend these methods into the domains of perception and visual short term memory.