This week, Journal of Neuroscience published the Journal Club commentary I’ve written with Tom Pfeffer, in which we discuss this paper by Simon Kelly and Redmond O’Connell. In short, we weigh the pros and cons about an electrophysiological signature of evidence accumulation in humans they claim to have found, and discuss what such a signature should look like in the brain.
Why is this interesting and important? The study of perceptual decision-making deals with the way in which our brain transforms incoming sensory information into an action. This is something you do all day, every day: you check the street for cars to decide if you can safely cross the road, look at a crowd at people and try and recognise your friend before waving at them, or browse through the menu at a restaurant to order the dish you’ll enjoy most. Mathematical models have shown that these types of decisions lead to the best results when we accumulate information over some time (say, a few seconds – an eternity for the visual system, which can recognize simple scenes in as little as a fifth of a second). For example, you will need to have a few images of the moving car on your retina to accurately judge how fast it is going. Once your brain decides it has enough information (also described as the crossing of a decision bound), you stop harvesting information and start to walk.
Steve Fleming has written a beautiful piece in Aeon magazine, Hesitate! Quick decision-making might seem bold, but the agony of indecision is your brain’s way of making a better choice, describing why the accumulation of information is so important, and how it helps us to make better decisions. Knowing more about the mechanisms underlying perceptual decisions can help us understand where they might go wrong, and how we can improve them.
Enter the brain. Although many great experiments in behavioural economics and cognitive psychology give us insight into the mental processes underlying decision-making, they give us very few datapoints per decision: the decision outcome itself, the speed at which it was made, and possibly how confident the observer is in their choice. If, on the other hand, one wants to investigate how perceptual decisions unfold on a millisecond time scale – from the moment of looking at an image to pressing a button to record the choice, we need to track what happens in the brain.
In monkeys, direct recordings of neurons have revealed that specific areas of the cortex seem to be doing exactly this – accumulating information until there is enough if it to make a decision. However, training a monkey to do such tasks is cumbersome, and sticking electrodes in their brains is a delicate art. We’d like to find if we can find such a neural signal in normal, healthy humans without opening their skulls – which will still enable us to measure decision-making as it happens. Recording EEG or MEG (basically, sensors that record the electric or magnetic field outside your head, generated by all your neurons communicating with each other through tiny spikes of electricity) can tell us very high precision what the brain is doing when (and to a lesser degree, where) in the seconds from vision to choice.
Unfortunately, however, linking a precise neural signal to the process of decision-making is quite difficult. Several papers have proposed different neural signatures of evidence accumulation, but all signals are not alike: some show up in motor cortex and predict which hand you will use to press a button, others are more related to visual information it self and don’t depend so much on how finally report your decision. These might seem small differences, but are crucial to the way we define what a neural signature of decision making should look like, and what we can do with it once we establish how easy it is to measure. Recently, a new paper was published about one such signature, which looks very promising but also has some drawbacks. For all the details, check out our JoN article!
CoCoSys lab 