Crime, the Hippocampus—and the Lingering Eye

Let’s say you’re in a crowded bar when somebody suddenly shoots at a patron. You clearly see a man carrying a firearm, but all hell breaks loose as you and everybody else scramble for the exits. In the terrifying seconds following the crime, you lose track of who discharged the firearm: it could have been 1 of 3 suspects. Afterward, the police interview you, but it is hopeless. Even bringing in the suspects for a lineup isn’t going to help you recall. There will be no “Perry Mason” moments, when the perpetrator breaks down under the weight of guilt and confesses to the crime. How can the authorities make an arrest?

They will be in for a tough time. Law enforcement officials know better than anybody about the long and storied literature concerning false memories, and the extraordinary unreliability of eyewitness accounts. People make things up all the time in high-pressure situations—seeing things that weren’t there, omitting things that were. The brain obviously records the information as it occurs in real time. But accuracy becomes only one version when the brain makes a conscious effort to retrieve the events.

What if there were a technological way to distinguish between these 2 versions of the same experience . . . the event that actually happened, and the version you perceived to have happened? The topic of this month’s column is an amazing result that promises to make this distinction possible. Using a combination of eye-tracking devices and noninvasive imaging, a group of researchers have uncovered an interesting way to get at a more accurate version of an event, even when conscious retrieval breaks down.

To talk about this extraordinary discovery, we will first need to review a few processes regarding human memory and some of the neural substrates that undergird it. Feel free to migrate to the “The results” section if words like “explicit memory” and “relational memory” processing are working parts of your vocabulary.

Thinking about memory

Present-day researchers use many ways to categorize human memory, some of which are clearly contradictory. That’s because there are many types of memory systems and subsystems, and not everybody agrees on just what they are. Is learning to ride a bike different from learning a foreign language? If a memory involves an emotionally competent stimulus, is that memory qualitatively different from the memory of a boring list of dead monarchs in a history class?

To make matters more complicated, many of these memory systems work in a semi-independent fashion. From a research perspective, the only clear thing you can say is that memory is not yet a unitary phenomenon.

One popular way to classify human memory abilities is based on whether the memory requires conscious awareness for retrieval. This idea was eventually transformed into a classification system involving so-called declarative memories and non-declarative memories. Declarative memories involve information you can physically declare—for example, that Thomas Jefferson was the third president of the United States. Non-declarative memories involve information you can’t declare—such as the learned ability to ski. Declarative memories are often called “explicit” memories; non-declarative memories are often called “implicit” memories.

The reason this classification system is so convenient is that there are distinct neural mechanisms and subsystems that underlie their function. Declarative memories are considered to be the province of the crown jewel of the medial temporal lobe—the hippocampus. The organ is ground-zero for helping to convert short-term declarative memory traces into long-term information. Damage to the hippocampus can lead to a selective inability to retrieve conscious events and facts while leaving non-declarative memory traces virtually intact.

Of course not everyone agrees with this rubric—the view is necessarily simplistic and has had recent empirical challenges. Certain patients with hippocampal damage possess deficits that are also associated with implicit memory formation. Some researchers believe the hippocampus is involved in yet another hypothesized memory category, so-called relational memory. This is the memory for the associations between discrete elements in an individual experience. Research findings in some labs implicate the hippocampus in establishing this type of memory too, whether a subject is aware of the experience or not.

It is also clear that other regions support the hippocampus in its quest to mediate memory formation, regardless of the category chosen. The prefrontal cortex (PFC), which is involved in decision making, planning, and foresight, clearly has a say in what will be encoded, stored, and ultimately retrieved. It communicates with the hippocampus during memory formation.

So what are we going to do? Even given these complexities, clear progress has been made in discerning the relationship between awareness and memory. Though it doesn’t settle the definitional issues, the data presented here show real promise of being practically helpful in a wide variety of settings. There may even be some real practical benefit to the law enforcement community.

The results

The inaugural idea of the project about to be discussed here was to explore whether the hippocampus could actually support indirect expressions of memories when conscious retrieval was failing. There were 2 machines involved in the experimental protocol. The first was a functional MRI, with special focus on images derived from the PFC and the hippocampus. The second was a tracking device that could detect eye movements and could determine where—and for how long—experimental subjects lingered over a given object in their visual fields. The eye-tracking device measured what commentators call REMEs—relational eye movement effects.

With these 2 machines calibrated and ready to work, the following 3-step experiment was deployed:

1. Healthy volunteers were shown pictures of individual faces superimposed on individual scenes (outdoor landscapes). The experiment was reminiscent of how animators draw characters on transparencies (cels), and then place the cels on previously painted backgrounds when recording for film.

2. After a delay, these volunteers were then presented with one of the previously encountered background scenes—sans face—and were asked to recall which face had been previously superimposed on it.

3. Finally, subjects were presented with a choice of 3 faces, and asked to choose which face actually went with a particular background scene in question. There was only 1 correct answer (2 of the 3 faces were distractors).

At each point, the subjects’ eye movements were tracked at the same time their brains were being imaged in the critical 2 seconds after the completion of the conscious retrieval task. The researchers were looking for:

• What the hippocampus did un-der conditions in which subjects matched the correct face with the correct background.

• What the hippocampus did when the subjects didn’t correctly match the items.

• What the REMEs did under both conditions.

What they found was extraordinary (Figure).

Extraordinary findings

Here are the results obtained when subjects were presented with the choice of 3 faces and correctly identified which face went with the background being studied:

• Activity in the hippocampus increased.

• Activity between the hippocampus and the PFC also increased.

• The eyes spent the most time lingering on the correct face (REME scores increased). There was a delay of 500 to 700 milliseconds between hippocampal activation and eye movement.

Here’s what happened when the subjects were presented with the choice of 3 faces and incorrectly identified which face went with the background being studied:

• Activity in the hippocampus increased.

• Activity between the hippocampus and the PFC decreased.

• The eyes spent the most time lingering on the correct face—even though the subject consciously gave a wrong answer.

The REME scores still increased. The eyes appeared to “know” what the correct answer should be, even though the subject did not consciously return the correct answer. By “know,” I really mean that the neural regions were directing the eyes. At some level, the brain was aware of the correct answer all along. For whatever reason, the organ did not return the correct answer to the subject. It means that accurate memory retriev-al was evident in eye movement behavior even when the subject’s judgment was incorrect.

This is a bombshell of a finding. Even if the subject was unable to recall a learned relationship between the face and the background, it appeared that the eyes and the hippocampus retained some of the information. And the eyes still responded correctly. At the very least, these data demonstrate that something beyond the hippocampus is required to make a previously encountered memory conscious.

These results suggest other ways to measure hippocampal memory formation that don’t depend on the subject’s conscious reporting. What other brain regions might be needed? The clue may come from the differential communication with the PFC. The connections between cortex and hippocampus became far less lively when the subject made a mistake. Because the PFC is involved in decision-making processes, its recruitment is hardly a surprising find. The intriguing decrease was not expected, however. It suggests new experimental directions in attempting to understand what the brain goes through when mistakes are being made.

One has to be careful of data like these, of course. One factor not controlled for in these studies is whether the subjects were conscious of the correct selection at the same time their eyes lingered on the matching face. They may have been aware of the correct choice, but in the intervening brief time between seeing the picture and selecting a face, they second-guessed themselves.

That’s hardly a deal killer. The real contribution of this work is to show that there are other ways of retrieving accurate information than direct interrogation of the witness.

And that has some powerful implications, especially for law enforcement professionals. Hooked up to these machines, you as a witness may indeed be confronted with a lineup of suspects who may have fired the gun in the bar. Your conscious recollection is just as fuzzy as ever, and you report that you cannot positively ID anybody. Your eyes (unbeknown to you) are consistently drawn to the actual perpetrator, however, revealing the correct version of the event. The machine detects this and the perpetrator, in the face of overwhelming technological superiority, breaks down and confesses. Sound like a fantasy? In the future, that may only be true of the perpetrator’s reaction.

This article originally appeared in the February 2010 issue of the Psychiatric Times.