Trying to understand the human brain usually leads to more questions. Will computer-based models be able to provide us with answers?
How does the brain work? How does it make sense of what we observe? Why do we sometimes hallucinate? Questions like these have stumped us for generations. The good news is researchers who study how the brain supports the mind are chipping away at our ignorance.
Using tools of science like the Human Connectome Project and functional magnetic resonance imaging (fMRI), they’ve discovered the three-dimensional organization of the living human brain. We now know quite a bit about which areas of the brain help us to perform certain tasks and where things go wrong in some forms of brain damage and disease.
What we’ve learned has helped one cognitive neuroscientist from The University of Texas Health Science Center at Houston (UTHealth) begin building a computer model of the human brain — one so faithful to the original that it will even make human errors in judgment.
To explore mysterious territory, it helps to have a map. After decades of study with autopsies, microscopes, imaging technologies and surgery, neuroscientists have created a type of map of our brain activity.
For example, we now know brain cells that deal with sensation and movement cluster in predictable areas. Touch your leg, and nerve cells fire in a specific neighborhood on the side of your brain called the parietal lobe — an event that can be seen on an MRI. We also know some body parts are allotted more brain space than others. In humans, the thumb, lips and tongue process much more sense information than, say, the knee, so their corresponding brain areas are larger, too. The proportions vary from person to person: In a violinist, extra brain space is devoted to processing sensory information from the left hand.
In the 1930s, neurologist Wilder Penfield, MD, mapped out these sensory and motor brain areas by stimulating them in patients with epilepsy. Against the backdrop of the parietal lobe, he was able to draw a distorted human form (big lips, small legs and so on) he called the homunculus. More recently, the Human Connectome Project is marshaling vast amounts of data to map the circuits that connect different areas of the brain.
Our understanding of neuroanatomy helps us predict brain damage based on a person’s symptoms, as well as predict symptoms from damage we see on a brain scan. The ability to produce language, for example, is localized to a region near the front of the left brain called Broca’s area. When a person with risk factors for stroke shows up in the ER able to speak only a few disjointed words, doctors suspect damage to Broca’s area, usually due to a localized blockage that deprives cells of oxygen. Interestingly, it doesn’t matter whether you speak or sign: Sign-language users who suffer strokes in that area lose their signing fluency, too.
Learning from the troubled brain
Brain maps show us other striking correlations between disease and anatomy. People with post-traumatic stress disorder (PTSD) show changes in the amygdala, an almond-shaped nugget of the brain that is thought to be involved in processing fear, anxiety and emotional memories. A person suffering from obsessive-compulsive disorder may have something wrong in the caudate nucleus, a part of the brain that helps a person note when a problem has been resolved.
More complex thought defies precise mapping. We don’t yet know if there’s an exact equivalent of the homunculus for memory, reasoning, decision-making, problem-solving, religious thinking or morality. These types of cognition probably involve multiple areas of the brain, and the way those areas activate may vary with context. It’s not always clear that common terms correlate to a single brain activity, either. Remembering the times tables versus a loved one’s face may light up different groups of brain cells on an fMRI, though we may call the activity “memory” in both cases.
Still, we’re gaining more understanding of which areas of the brain seem to support these types of thought. For example, the hippocampus, an oblong region near the center of the brain, plays an important role in learning and memory. Trouble forming new memories, as in Alzheimer’s dementia, can result from damage there. The hippocampus also plays an important role for sufferers of Huntington’s disease, an inherited disorder that leads to both movement problems and dementia. These patients seem to diverge from healthy people by the failure to create new neurons in that region.
Decision-making, which involves figuring out one’s choices and weighing consequences, seems to require the help of the prefrontal cortex, an area behind the forehead that connects to many other regions of the brain. That same region is important for behavior and emotion, as we learned from the famous case of Phineas Gage, a 19th-century railroad worker whose personality turned erratic after he survived a freak accident that drove an iron pole through his left frontal lobe.
Biases and ‘sensemaking’
Even without disease or injury, thinking routinely goes wrong — or at least veers from what we generally consider to be rational. We’re all subject to cognitive biases every day, and there are a host of them. For example, we cherry-pick information that supports our argument (confirmation bias); we unduly base our decisions on the first piece of information we happen to receive (anchoring bias and the order effect); we underestimate how long it will take to complete a task (planning fallacy); and we stereotype people — one of the oldest and ugliest errors.
Understanding these biases is central to understanding human thought, says Hongbin Wang, PhD, professor at UTHealth School of Biomedical Informatics. Wang is helping to build a computer-based model of the human brain that will perform the task of sensemaking in the way humans do.
Imagine looking at a series of images: a car driving down the street, an explosion in a building, a map showing a cluster of cell-phone calls. How do you analyze these seemingly random events and figure out what happened? What kinds of irrational cognitive biases will sway you — for example, will you assume without evidence that the car was involved with the explosion, or that the explosion was due to a bomb rather than, say, a gas line?
That’s sensemaking, Wang explains. It’s how humans generate and evaluate explanations for information that is sparse, noisy and uncertain. The process involves perception, attention, learning, memory and reasoning — it’s so complex, in fact, that modeling it means modeling the whole brain.
Wang’s computer model will examine a group of images like the ones mentioned above and then pick an explanation for those images. (For comparison, human subjects have field-tested the same scenarios.) And because human sensemaking involves bias, Wang and his colleagues are building their model to make the same predictable errors that human beings do. “Our model needs to be able to reproduce those biases and give an exact detailed explanation of why in this situation those cognitive biases happen,” Wang explains.
It’s a big undertaking. For sensemaking, our brains rely on activation in the visual cortex, the parietal lobe, the prefrontal cortex and so on. Each of these brain lobes could be thought of as a single computer that runs many “programs.” Similarly, Wang is collaborating with colleagues in California, Colorado and Pennsylvania; each of them is building one part of the model. The components, like the parts of the brain, will be linked together later using a supercomputer in Colorado.
Wang’s model may improve our understanding of decision-making and cognitive biases during sensemaking, he explains, which may make it possible to develop an intelligent tutoring system that can help to improve human performance.
Even if Wang’s model succeeds in simulating sensemaking, we’ll be left with plenty of tantalizing questions about the brain and its relationship to the mind. What is the language of thought? Is it words? Symbols? And the mystery of consciousness is a nut no one has cracked.
The pace of discovery may soon speed up. In April 2013, President Barack Obama launched the BRAIN (Brain Research through Advancing Innovative Neurotechnologies) Initiative, an effort to invest in technologies that will help us watch the brain in action and advance our understanding of thought, learning and memory.
Still, no one theory, model or technology is going to answer the big questions. “The brain needs to be understood from multiple levels,” says Wang. “But given how much we know about the human genome, how much we know about cognitive function, advances in computer science and information science — there is no better time than right now to tackle this issue.”