Researchers Know What You Were About to Say; fMRI Used to Detect Memory Storage and Retrieval

facebook twitter google print email
Media Contact:Greg Lester | glester@pobox.upenn.edu | 215-573-6604December 22, 2005

Using functional magnetic resonance imaging, or fMRI, researchers at the University of Pennsylvania and Princeton University have provided evidence that the act of recalling a memory is a bit like mental time travel.  Their study, presented in the Dec. 23 edition of the journal Science, demonstrates that the same areas of the brain that are active during an event are activated when a person attempts to recall that event seconds before the memory surfaces.

"This study shows that, as you search for memories of a particular event, your brain state progressively comes to resemble the state it was in when you initially experienced the event," said Sean Polyn, a post-doctoral fellow at the Computational Memory Lab in Penn's Department of Psychology.  "It is all part of the brain's ability to cross-reference memories, pulling together separate pieces of information from an elaborate network of stored representations to recreate an event."

The inability to recall something is a common frustration.

"An every-day strategy for getting at lost memories involves using a part of a memory to pull out the entire thought, much like when you try remember where you put your keys last night," Polyn said.  "If you recall that you were washing dishes, that might trigger associated memories, leading you to remember that your keys are next to the sink.  We refer to this phenomenon as 'bootstrapping.'"

According to Polyn, this "bootstrapping" effect occurs as a brain area called the hippocampus helps sort through the storage bins of memory, returning the brain to its state at the time of the initial experience.  Polyn believes that the knowledge of how the brain uses its memories could be applied to designing more detailed models of memory, which could help treat brain disorders such as Alzheimer's disease and epilepsy.  This knowledge could also guide creation of more self-sufficient artificial neural networks and robots.

The study also broke new ground in how fMRI could be used to study how brain activity changes from second to second.   Polyn and his Princeton colleagues gave participants 90 things to remember divided among celebrity faces, common objects and famous locations using the fMRI to detect which parts of the brain were involved in the learning process for each category.  They developed a technique that could track the brain activity corresponding to each of these categories as participants retrieved memories. As they remembered, the technique provided a second-by-second readout of how the brain searched for information.

So participants would not feel compelled to "cram" the 90 items, the researchers presented them in the form of a series of judgments, for example, asking whether or not the subject liked or disliked a labeled photograph of comedian Carrot Top.  These judgments were designed to make participants think more deeply about the items, allowing them to form stronger memories.  These judgments were interspersed with simple arithmetic questions to keep the participants from simply memorizing the items.  The subjects were then asked to freely recall the 90 items, in whatever order they could remember them.

As the research team reviewed the data, they could see how the portions of the brain that stored memories of faces, for example, would activate several seconds before the participant began to name the celebrities.  According to Polyn, objects, faces and locations are all stored differently in the brain.  When the participants moved from one category to another, the researchers noted a corresponding shift in brain activity.

"The results of this experiment suggest that when we think back to the past, each detail we remember triggers another, until the memory returns completely," Polyn said.  "In that sense, memory retrieval is like revisiting the past; brain patterns that are long gone can be revived by the memory system."

###

The research was carried out in the Princeton laboratory of Kenneth A. Norman.  Polyn's other Princeton collaborators were Vaidehi S. Natu and Jonathan D. Cohen.  Funding was provided through National Institute of Mental Health grants.