Mind Matters

Mind Matters is Sciam.com's "seminar blog" on the sciences of mind and brain. Each week, top researchers describe their disciplines' most significant new findings -- and what they, as fellow researchers, find most exciting, maddening, significant, odd, or otherwise noteworthy in the research driving their fields. Blog visitors can participate. We hope you'll join us.

Do you need a hippocampus to imagine a museum?

by David Dobbs

As we explored in the very first Mind Matters post, neuroscientists everywhere agree that the hippocampus is crucial to memory -- but have rich and interesting disagreements about how this brain area creates and manages memory and what roles it might play in cognition. This debate was freshly enlivened in early 2007 when an innovative paper by Demis Hassabis (a former chess prodigy and games designer) and colleagues at the renowned University College London lab of Eleanor Maguire proposed that the hippocampus is vital not just for memory but also for imagination. As hippocampal researcher Andre Fenton notes in his review below, this discovery suggests both a vital new role for the hippocampus and a narrative-building mechanism common to memory, imagination, and thought. Interesting new ground, Fenton finds, but not without its hazards.


by Andre A. Fenton

The Robert F. Furchgott Center
for Neural and Behavioral Science
State University of New York, Downstate Medical Center
Brooklyn, New York

"Imagine there's no heaven..." John Lennon challenged us in his 1971 song "Imagine." Apparently that could be hard if you don't engage your hippocampus.

At least that is what Demis Hassabis and fellow researchers at University College London suggest in their paper, "Patients with hippocampal amnesia cannot imagine new experiences."

This study is the first to examine objectively the neuropsychology of imagining events -- a process known in neurocognitive circles as "construction." The study asserts that imagining events requires some of the same operations we use to remember events: a sense of subjective presence in the scene, for instance, along with the use of imagery and the selective retrieval of relevant information from factual memory. (Factual, or "semantic" memory, refers to recall of discrete facts, such as "bicycles have wheels." In contrast, "episodic" memory is of actual autobiographical events, such as the bike ride you took last week.) The study also postulates that construction requires awareness of an imagined experience's narrative, temporal and spatial organization. Hassabis and his colleagues argue that these organizational elements, which are known to be fundamental to recollecting episodic information from autobiographical memory, are also needed to imagine events.

New Job?

The paper also breaks ground in hypothesizing that mental construction -- a form of thinking, really -- is a key operation of the hippocampus. This idea gives the much-studied hippocampus a new role that it is not granted in the two main theories of hippocampal function. The traditional, "relational memory" or "memory bank" view, based mainly on primate experiments, asserts that the hippocampus's main job is storing declarative, or explicit information (that is, memory we can recall and tell about) -- especially episodic memory, the autobiographical events that make up our own remembered lives. An alternative view of hippocampal function, the "cognitive map" theory, is based predominantly on rodent experiments. This theory asserts that the hippocampus organizes remembered experience mainly by storing it in a neural "cognitive map" of the physical world. In short, the more traditional "memory bank" theory emphasizes the storage of sequences of events in time and space -- I shopped for pants in Soho on a hot Saturday afternoon, the sales clerk flirted with me, then, feeling handsome, I got a haircut nearby and enjoyed the air conditioning there. The cognitive map theory, by contrast, emphasizes the storage of sequences in space -- I visited three places along Spring Street, then crossed Broadway to get to the barber's.

er of these established theories, however, has much to say about construction of events once the factual information in memory is stored and accessible to the imagination. But here Hassabis and his colleagues claim the hippocampus plays an important role in mental construction even without explicit reliance on episodic memory, and that this construction is a crucial but overlooked component of both memory recall and imagination.

Put Yourself in a Museum

Hassabis and colleagues studied five patients with amnesia caused by hippocampal damage. (These patients, like most with significant hippocampal damage, have difficulties forming new memories.) The patients, as well as some people without damage used as control subjects, were asked to imagine new experiences such as standing in the midst of a museum or street market. The study's main finding is that the amnesia patients did less well at imagining events than the controls did. However, it was hardly the case that the patients could not imagine new experiences, contrary to the paper's title. In fact one patient could imagine as well as the control subjects could.

At this point I must warn the reader of my bias. My research focuses on the electrical signals generated by hippocampus neurons as a rat solves formally complex spatial problems. I imagine that my collaborators and I are working towards understanding the neurophysiological basis of thinking -- though I am not even sure that rats think, in the sense of flexibly pondering and solving problems. (I suspect most of my colleagues are even more doubtful about rat cognition.) We measure physiological signals that might indicate rat thinking. So far, however, there have always been other, simpler explanations of what might be causing those signals. This reasonable skepticism about the explanation for these signals makes me similarly skeptical about the Hassibis results: Apart from impaired imagination, are there simpler possible explanations for the impoverished imaginative responses of the amnesia patients?

We can safely exclude factors such as IQ, education, age and the like, because Hassabis made sure these variables were similar in patients and controls, as was the ability, we are told, to access factual (semantic) information. In addition, based on the subjects' self-assessments, Hassabis concluded that comprehension of the instructions, the self-perceived task difficulty, and such things were equivalent in the amnestic patients and controls.

Unwelcome Memories

I find less convincing, however, the experiment's control of another potential cause for differences in performance, which is that control subjects might have been recalling actual experiences, rather than imagining new ones, as they attempted to "imagine" their new experiences. To prevent this recall from happening, the experimenters asked the subjects to imagine commonplace scenarios and not to use personal memories. But doing so is easier said than done -- and very hard to verify.

My initial doubts about that control grew when I tried to construct being in an imagined museum. In constructing this scene, I found myself relying heavily on my own recollected experiences. I drew on memories of the National Museum of Natural History in Washington, D.C., to construct my imagined museum's main hall, the National Museum in Prague for the artifacts, the Tate Modern in London for my sense of the people, and the National Gallery of Canada in Ottawa for my sense of the light. I admit it may be a character defect, but I constructed my imagined experience with facts and features culled from recollections of real experiences. This construction was within the letter of the Hassabis experiment's rules but contrary to its spirit. Had I been in the study, I could in good faith have reported that my imagined experience was not a recollection of actual experience. Yet I clearly drew on episodic memory. And amnesia patients have severely impaired episodic recollection -- indeed, that's hippocampal amnesia's hallmark. Could this primary deficit have made fewer facts accessible to fuel their imagination? The researchers argue that they ruled out this possibility because semantic memory was equivalent in the patients and controls and because the subjects said they didn't draw on episodic memory. But I'm not convinced.

There are more certain ways to rule out a basic difficulty to assess facts, and the authors did try one of them. They developed an "assisted" variant of the construction task in which they provided relevant photographs, sounds and smells throughout the test, presumably reducing the need for factual recall. Unfortunately, only one subject (a patient) was given this control test. The assistance did not improve his performance. But before concluding that these patients "cannot imagine new experiences" I would expect each subject to get the factual assistance test variant. After all, one patient was unique in that he had normal performance. Could the subject that received the assistance also have been unique?

The study's central finding would have been substantially bolstered if it had been confirmed by an additional test. (Such additional tests are standard practice in animal research.) For instance, to test whether these patients have impaired imaginations independent of their access to facts, the authors could have used a dollhouse-style model or a computer simulation of a museum, market, or fictitious planet and then instruct the subjects to imagine being a character in that fictional world. That would automatically provide all subjects with the same factual base and test their abilities to construct new imagined scenarios.

These caveats concern completeness. As a start, however, this imaginatively conceived paper breaks interesting new ground, and if its findings are expanded and upheld it could initiate a new direction in thinking about hippocampal function as a structure that "constructs" memory and imagination in addition to its recognized role in storage. I want the hippocampus to be crucial to imagining. But without adequate experiments I can only accept a more parsimonious account for the findings here, which is that the hippocampus, as we already know, is crucial for recalling facts from experience. Andre Fenton is an assistant professor of physiology and pharmacology at the State University of New York Downstate Medical Center, where he studies the neurophysiology of spatial cognition and memory.

-- Edited by David Dobbs at 01/02/2008 8:37 AM