Computers Knowing How Humans are Remembering

Standing in the grocery store, trying to remember the shopping list left at home, our brains employ a variety of memory systems to assist in the recall process. Now researchers are finding that they can use computers to model these same memory systems humans use to remember an event.

When the brain is attempting to remember an event, it is retrieving from memory not just that single event, but also everything surrounding that event, the context. By studying the contexts applied in remembering, the computer models are able to replicate human data, according to the study published in Neuropsychologia in January 2012.

According to Ehren Newman, a memory researcher at Boston University, “Computational models provide a means to understand un-intuitive patterns in the data.” By better understanding human memory systems, there are hopes that researchers will be able to hone in on understanding specific aspects of memory.

One such aspect includes tracking changes in memory retrieval as our brains age. As we age, our ability to recall memories generally decreases, but with this computer model it may be possible to identify which memory systems are not being employed as they were in younger years. Looking at younger and older memory data for a patient allows researchers to compare what changes in brain activity create the memory deficiencies. Understanding and being able to predict typical memory decline may lead to being able to identify unusual decline and use that identification to predict potential memory-related illnesses.

In addition to using this understanding to see how our memory systems change over time, this information could be used to see how the recall process is different for schizophrenic or amnesic patients, and address the cognitive problems associated with those illnesses.

“Many researchers are looking at how the brain is organizing recall underlying the memory process,” says researcher Joshua McCluey, “But not many are looking at free recall with an fMRI.” By looking at blood flow to the brain, an fMRI makes it possible to look more closely at what the brain is doing and monitor neural activity, which leads to better results. Because the results found are for free recall, which is when a participant remembers items in no particular order, the data gathered is more complex and representative of human experience.

To better understand how memory works and hopes to apply this understanding to issues like memory loss, schizophrenia, and amnesia, researchers at Vanderbilt University are studying the brain both when an event occurs and when it retrieves that same event from memory. This data is then used to develop a computer model of human memory.

The researchers generated memories for each participant by having her read a list of 12 words and mentally decide whether the object would fit in a shoebox or mentally decide whether the object is alive for each item on the list. Making this judgment serves to help solidify the memory in the brain. Meanwhile, an fMRI machine scanned each participant’s brain to see how it stored those memories. Later each participant was asked to recall every item on the list to the best of her ability. Again, during this recall portion of the test, the brain was scanned to see how the memories were retrieved.

Patterns that were observed and modeled by the computer include recency effect, which shows memory preference for items at the end of the list, the items most recently read. They also found a contiguity pattern, where participants are more likely to remember the word appearing immediately around, especially after, the previously recalled word. When they applied all the different patterns they discovered to the computer model, they were able to see that the computer, with good accuracy, could replicate the way that the human brain remembers events.

While this computer model can duplicate human memory, it cannot serve as a substitute for human memory and won’t help you suddenly remember those grocery items you’ve forgotten to purchase. It may, however, serve as a way to better understand what goes into making and recalling memories as researchers try to wrap their minds around the brain and how it functions.  

Finding the sources

So for my last post, I had quite an experience finding the necessary sources. I ended up talking to Dr. Lukehart, the primary research contact I had, because he was my professor for Nanochemistry, which was one of my favorite Vanderbilt science classes and a large part of that was his excitement about the topic. I found a paper he had recently published, and decided to talk to him about it. He was so easy to talk to and ask questions of, and gave me the lock and key metaphor I was really proud of in the article.

For my second source, I had a bit more trouble. I emailed Dr. Wright, another one of my professors at Vanderbilt. He taught BioInorganic Chemistry, which is exactly where these nano-particles could be studied. They are used in human life (bio), they are metal (inorganic) and they are used by reactions (chemistry). The article was due Friday at 6, and 3 pm rolled around that day and he still hadn't gotten back to me. I decided to go to his office in person to find him and get a quote. As soon as I stepped off the elevator in the Chemistry building on campus, I saw him in a conference room talking to his graduate students. I waited outside the room for him, then as he came out got a chance to walk and talk with him to his office as he told me what he thought was important about the nanoparticles.

For my final source, Dr. Hallahan, I emailed him about finding a chance to talk, and his graduate student responded to me, and we set up a time for me to call. Unfortunately, the night before the phone call I found out that I would be on a roof, nailing up shingles at the same time our phone call was scheduled. For fall break I went on a service trip to Knoxville, and I accidentally figured the time change in the wrong direction. Oops. So I sent a quick email apologizing for the change of plans and a few questions I was going to ask in the hopes that he would still respond over email. Lucky for me, he did, and I got the information I needed to finish the article in time for my second deadline.

Finding sources was definitely the most difficult part of writing this article, but I can see how it paid off in the final product. I'm pretty proud of my little news article.

Two-for-one: locate and kill cancer with a single agent

Scientists may have found a way to more accurately locate cancer cells and, at the same time, destroy them. New, nano-sized metal particles can bind to tumor cells, emit a fluorescent glow, and then be heated with a magnetic field to kill nearby cancer cells. By uniting the process of imaging and treatment, therapy becomes more effective and targeted to the cancerous cells.

There are numerous methods of cancer imaging, many of which involve using an identifiable molecule that fits into a cancer cell receptor like a key fits into a lock. This process takes some time (in this study 48 hours), and requires that the molecule does indeed fit into the receptor of a cancerous cell, that the molecule can be distinguished during imaging, and that enough of those molecules are located in close proximity to recognize an increased concentration.

For years, researchers have suspected that nanoparticles are good candidates to improve cancer imaging because of the likelihood that more of them would bind to a tumor. In 2010 Dr. Lukehart, a Chemistry professor at Vanderbilt, showed that this was true by using nanoparticles made of iron and platinum. Each nanoparticle was covered with thousands of peptides,  molecules used to bind to a cancer cell. When the tumor is hit with radiation, the nanoparticles bound to cancer cells fluoresce, giving off light waves not visible to the human eye. A machine able to detect these waves is used to locate where they are coming from, to create an image of where in the body cancer cells are. The image created using nanoparticles was much brighter and easier to analyze than when single molecules were previously used. This improvement is because  when a peptide binds to a receptor, like a key fitting into a lock, there are thousands of metal molecules within the nanoparticle it is attached to. Since each metal molecule is giving off the light waves, the machine is able to detect thousands more light rays coming from each location where a peptide is bound to a cancer cell.

Researchers outside Dr. Lukehart's lab see the advantage of this technique. “This technology is cool,” says Dr. Wright, a bio-inorganic chemist at Vanderbilt who did not participate in the research,“ It let’s us see how well radiation is going.”

Using these nanoparticles to image cancer in the future is likely, not just because of the superior images that result, but also because of the possibilities of treatment using these same nanoparticles. The iron and platinum nanoparticles used are paramagnetic, which means they line up in magnetic environments. By rapidly reversing the polarity of a magnetic field, the nanoparticles will create heat, which can kill the cells located nearby.

The platinum part of this nanoparticle introduces another way that this drug could be used to treat cancer. Platinum has been shown to be an anti-cancer chemical. In fact, the anti-tumor properties of the element are already in use in the platinum-containing chemotherapy drug, Cisplatin, used to treat testicular cancer. By using nanoparticles that both contain platinum and are surrounded by peptides which link to cancer cells, as was done in this study, it is possible that the simple introduction of this imaging agent into the body could serve to treat cancer without any external influence.

Dr. Hallahan, head of the Department of Radiation Oncology at Washington University School of Medicine in St. Louis, suggests, “These nanoparticles will first be used for patients with poor prognosis cancers. These include patients with recurrent disease and patients with [inoperable] cancers.” Treatment with these nanoparticles fills an unmet medical need, as there are “no treatment options for most of these patients.”

In this particular study, only the imaging abilities of nanoparticles were analyzed and only lung cancer was investigated. According to Lukehart, however, this method could be easily adapted for other types of cancer by simply changing the peptide linker to one that binds to the targeted cancer cell.

The largest obstacle in the way of getting this imaging technique and treatment to patients is the creation of these nanoparticles. Scaling up their production  is not only difficult, but expensive, and could hold back the application. Lukehart hopes, however, that “if there are really hot results in future research, we may see it in place in only a couple years.”