Sunday 18 August 2013

What are experimental psychologists good for?

Just what are experimental psychologists good for? Or perhaps more specifically, what are experimental psychologists good for in the field of cognitive neuroscience? Although my non-psychologist colleagues haven’t been blatant enough to ask me this question directly, they have certainly alluded to it on many occasions. Indeed, experimental psychologists themselves sometimes suffer from self-esteem issues. A PhD student the other day commented to me that they had done psychology as an undergraduate, but felt they should have done a ‘harder’ subject. Apparently, it would have been easier to pick up psychology without formal tuition, compared to say brain anatomy or the physics of MRI or the complexities of synaptic transmission. To this I say poppycock.

Cognitive neuroscience is an inherently multidisciplinary field. Physicists, medics, computer scientists, mathematicians and, yes, experimental psychologists, rub shoulders with each other on a daily basis. This allows for research that would never be possible if people with such academic backgrounds were isolated. For example, carrying out complex analyses of functional neuroimaging data acquired in specific patient populations. The more mathematically minded are needed for pushing the boundaries of analysis techniques, medics are needed for the patient populations and psychologists are needed for experimental design and interpretation of the results. Few would dispute this, but a common assumption is that the ‘experimental design’ part is the easy bit. We’re interested in memory, so let’s give them the memory component from a neuropsychological battery, or a word list to memorise, and see what’s going on in the brain! Yay brains! Interestingly, this argument doesn’t seem to apply in reverse. Although many do it, you wouldn’t go around bragging that computer scientists and physicists aren’t needed because analysing neuroimaging data is the easy bit, you just use the default settings in SPM or FSL (or AFNI! Hello Americans!), click some buttons, and hey presto – blobs on brains!

So is experimental design easy, or is it easier to ‘pick up’ without formal tuition? You already know I’m going to answer with a resounding NO. Learning how to design good, well controlled, experiments is hard. It requires years of learning other people’s designs, designing your own experiments, getting formal feedback on those designs, running experiments, refining your previous designs and, importantly, pouring over data. It is only by getting your hands dirty designing, running and analysing experiments, within the structure of a formal supervision process, that this skill can be appropriately learned (actually, it is never 'learned', you are always learning). So the next time someone makes a back-handed remark about psychologists, stand up for yourself and your academic background. No-one else is going to!

There is a beauty to experimental design. Just as mathematicians see beauty in equations, I see beauty in experiments. I imagine it is much the same as the beauty an engineer sees in their structural designs, in that it is a mix of creativity and logic, of elegance and practical necessity. You start with a problem, or a research question, and map out the most obvious way to answer the question. You write it down. You look at what you’ve scribbled and instantly see an issue. You think. You come up with a solution. That introduces another unforeseen problem, so you change that. The whole thing starts to get a bit complicated. You have an epiphany and see a simpler means of getting at the same underlying issue. You tweak that further. You repeat over and over until you are as close to happy as you will get. This to me is the best bit of research in my field. The process excites me, as I get closer to the final design and the possibility of answering a previously unanswerable scientific question.

That is what experimental psychologists are good for in a multidisciplinary subject such as cognitive neuroscience. Clever experimental design, grounded in cognitive theory, that gets to the core of the question any research endeavour into the brain choses to ask.


  1. Great piece. I think often it is easier to just go with canned experiments and call it good because it lets the cog neurosci say their data apply to others.

    Unfortunately, as a mouse psychologist as I call myself, that leads to infinite water maze that gets analyzed in strikingly odd ways to make a point not at all related.

    As a biologist by education trained in experimental psych through undergrad--I can say that I am MOST proud of my task and when others adopt them. I hope others read this and see their psychology training with pride, not shame or embarrassment.

    1. Thanks Michael. I agree that taking an experiment 'off the shelf' can be dangerous in the wrong hands. There is a definite tension here though as we do want some methodological consistency across studies. For example (off the top of my head), when looking at the effects of particular drugs on hippocampal function, it is useful to have a standard 'hippocampal-dependent' task that can be used across labs. Experimenters still need to be aware of the limitations of whatever task they choose to use though - i.e., even if it's a standard task we should never just 'plug and play'.


  2. As Michael Saling used to tell us during clinical placement, we've become a sort of low self-esteem profession, in which hard-won knowledge is taken for granted and we assume that the knowledge and skills of a well-trained psychologist are common-place. Yet it only takes a cursory glance at some badly designed studies to quickly dispel that notion.

    Studying the manifold functions of the living brain is hard enough. It would be silly to take for granted what we've already gained by assuming anyone could just "pick it up" without so much as reading a basic text on the subject.

    1. I couldn't agree more (as you may have already guessed)!