Behavioural should not be behaviourist

What differentiates behavioural economics from psychology? A common answer is "we are interested in behaviour".

For example, psychologists studying group identity might use a questionnaire measure, asking subjects "how proud are you to be in your group"? Economists, instead, would test whether they gave more money to their in-group. "Real choices have costs," goes the slogan.

Costly choices are a great experimental tool. But I had an epiphany recently, when a psychologist remarked that economists are using a theory psychologists gave up fifty years ago – behaviourism, the idea that you could ignore the mind and just study how stimuli affected behaviour.

Only studying behaviour makes sense as long as the mind has no "state". If stimulus X always causes behaviour Y, then we only need study Y.  But of course the mind has state: for example, our choices depend on our moods.

We need to study these states directly, so as to improve our theories by clarifying the links in their causal chains. Why are subjects more selfish if the game they are playing is labelled the "Wall Street game" rather than the "community game"? Maybe the market-oriented phrase puts subjects into a materially self-interested mindset. OK, so when does that happen in the real world? For an answer, we will need to measure this mindset directly and learn how it is induced.

This is doable. Yes, people may lie, or misinterpret their own behaviour; but psychologists have been devising valid, reliable questionnaire measures for decades. (FMRI data seems more easily accepted by economists than questionnaires, perhaps because it seems more like "hard" science – or because it is impressively expensive?) We can use these measures, or create similar ones of our own.

Without good theory, experimental economics risks becoming a pile of unorganized, uninterpretable results. For good theory, we need to open and study the black box of the mind.