A Surprising Use for Randomness in the Brain

If brains are like computers, how can our brains generate randomness?
— Timothy Myers (paraphrased)

When I first read this question, I wondered if our brains would benefit at all by generating randomness.

I found a probable answer last Friday as I stared in utter confusion at the 40 odd kinds of salsa stacked up in the grocery store. Five options remained even after applying my selection criteria. Braving it, I closed my eyes, pointed my finger randomly, and picked the jar closest to my hand. Although far from a perfect example, it managed to get me thinking perhaps decision-making is why our brains need randomness in the first place.

Let me say it right away, yes the brain does, in fact, generate randomness! We’ll talk about what randomness means in the brain - does it mean generating random numbers, much like a computer, or something more? We will then talk about why, when and finally, how the brain generates randomness.

The human brain does not do as well as a computer when asked to generate true random numbers. Randomness in the brain means something different – it is born from neurons that spike spontaneously or as a response to stimuli. It turns out that spiking behavior of neurons is very noisy, and somewhat unpredictable. Noise in a neuron is postulated to come from genetic, electrochemical and thermal variability, and from disturbances when neighboring neurons spike.

The Noisy Brain is an excellent book for the many theories of why and how our brains could use randomness, but decision-making is probably the best-studied one. Decision-making is a key brain output - this is made clear when we use the brain-computer metaphor. Our brains are far more complex than computers, but they both take external inputs and return an output. Our brain processes input in the form of sensory stimuli to generate output in the form of decision-making, usually resulting in body movement. For example, our nose feeds odor molecules to the brain to make a decision; and if it’s a bad smell, we scrunch up our noses and move away.

The requirement for randomness in decision-making presents itself quirkily through an old paradox about a hypothetical donkey, which would starve to death because of its inability to decide between two equidistant equally yummy bales of hay. Of course, it’s just an exaggerated version of my salsa problem. Our decisions are influenced by past experiences embedded in our memory. But when presented with choices that have somewhat equal outcomes, it is important to break symmetry to arrive at a decision. There is evolutionary advantage in being able to generate such randomness for decision-making — it allows the brain to take a detour from past experiences when they don’t provide a solution, and take a fresh unbiased perspective to navigate a new or confusing landscape or situation.

So now that we’ve seen some answers for why the brain needs randomness, lets see what scientists have to say about how the brain computes when it’s time to switch from experience-based decisions to random choices, and how such randomness is created.

Answer to the first question comes from a recent study done by Gowan Tervo et al in the lab of Alla Karpova, which explored the neural basis of random decision-making behavior in rats. In brief, the scientists presented rats with the choice of either turning left or right for a potential food reward. A computer recorded these choices and predicted the next choice. Rats received the food reward only if the prediction was wrong, so the rats had to beat the program to get the reward. When the program predicted weakly, rats mostly used past experience to decide left or right. But when a more competitive program was used, rats switched to random decision-making for the reward. This work is evidence that the brain uses randomness for decision-making when facing unpredictable or challenging situations.

Tervo and the team found that the switch from strategic (past experience-based) to random behavior lies in the anterior cingulate cortex, the part of the brain involved in decision-making and reward anticipation. Increasing the level of a stress hormone called norepinephrine in this brain region could activate random behavior, while decreasing it activated strategic behavior in the rats.

Needless to say, there can be a thing as too much randomness in the brain. The amount of norepinephrine was enough to predict if the rats would use their encoded understanding of the world or switch to random thinking. But rats in the study sometimes got ‘stuck’ in the random mode. They could be released from this state by suppressing the release of norepinephrine in the anterior cingulate cortex.

The answer to the second question – how does the brain create random behavior – is still a bit murky. There is some evidence that this is achieved by making use of the inherent random noise in neuronal spiking. Computational models predict that when faced with confusing choices - encoded as different neural pathways - the deadlock can be resolved in a probabilistic way so that the neural pathway (or choice) with the most spike noise would ‘win’. Essentially, this means that during decision-making behavior, inherent noise can be read as a signal by the brain. Behavioral tests for these computational models would hopefully unravel more insights into this process.

While our brains employ randomness for certain kinds of decision-making, researchers are getting busy employing randomness to diagnose mental illnesses. The Random Number Generation (RNG) task requires people to generate long sequences of numbers at random; and measures executive function of the brain (working memory, reasoning, decision-making, and problem solving). In general, people perform non-randomly at this task, but those with disorders such as schizophrenia show greater non-randomness in the RNG task compared to healthy subjects. Scientists are still working out variations of the RNG task and the precise parameters to quantify task output, which can be useful for diagnosis some day.

Randomness is fascinating, but what does it mean to have randomness? Could it mean that free will exists? Is an ‘out-of-the-box’ idea a brainchild of randomness - a result of noisy neurons talking to each other in a random way? Those are big questions for philosophers and scientists to work out. But while that happens, perhaps you could do an experiment of your own – check out if you can behave randomly – if you’ve got some time to kill!