Are there smartdrugs?

Bostrom,N (
Tue, 12 Oct 1999 16:51:09 +0100

I have been too busy to post anything recently -- to the point where I got an email today asking if I was dead. So I thought that I should give a life sign. Here are some ruminations on whether we can expect there to be smartdrugs. Comments?

Are there smartdrugs?

One theoretical argument against the possibility of smartdrugs is that if there were a simple way of boosting intelligence, evolution would already have discovered it.

Some things might still be possible. For example, if a higher brain metabolism would increase cognitive performance, it might still not have evolved because of the energy cost. So theoretically, there might be a drug
(caffeine?) that would make us think better by boosting mental energy. Other
examples might be if a drugs that facilitate some aspects of cognition at the expense of others. Say, it facilitates memory consolidation but inhibits ability to notice patterns and generalities (ampakines?). Or it promotes abstract thinking and reflection but at the cost of making us slightly absent-minded and less aware of our what we are not currently focusing on
(nicotine??). Such effects might not have benefited our evolutionary
ancestors, but they could nonetheless be useful for people living in today's society.

What one wouldn't expect if the above theoretical argument is correct is to find some drug that makes cognition more efficient without any downsides.

But I think the argument might be flawed. Consider the following model: there are many different variables that influence cognition (release-rates of neurotransmitters, density of certain types of synapses, concentrations of enzymes that break down the neurotransmitters etc.). Let's assume that each of these variables are normal distributed around some mean m in the population. Evolution predicts that m should be close to optimal - but for the population as whole, not necessarily for most individuals. For what is the optimal value of a variable in one individual might not be optimal in another individual. How much monoamideoxidase of type B you should have, for instance, presumably depends on how efficient your dopamine-reuptake channels are and how many dopaminergic neurons you have etc. etc., and this will be different in different individuals. So evolution might have a difficult time providing you with the optimum levels (for you) of MAO-b.

If this is correct, what one would predict is the following: We will not find smartdrugs that improve cognition in the population as a whole (except those that have some downside - like energy expenditure or shift of focus, as discussed above). What we may find, however, are drugs that improve cognition in some subjects and worsens it in others. Testing on a large group of people, we would not find any effect.

Such drugs could still be very useful, but they would require some individual experimentation. You have to try many things to see what works for you. A proper clinical trial for a proposed smartdrug could thus have two stages: the first stage identifies the people who seem to react most favorably, and the second stage tests the overall effect on this special population. My impression is that this is not usually done today, with the possible consequence that many smartdrugs have gone unnoticed. Thus one would predict that the smartdrugs that are discovered with today's methods are found in the context of treatment of some special disease. Maybe having that disease is correlated with having one of your brain variables displaced in a certain direction from its optimum. Maybe people with some kinds of brain damage typically benefit from higher levels of acetylcholine, for example, and this could lead to a drug like Piracetam being identified as a smartdrug. (The reality is more complicated, but it illustrates the idea.)

This reasoning only shows the theoretical possibility of smartdrugs. Whether any useful ones can actually be developed is of course an empirical question. Still, it makes the search look much more hopeful. And it suggests that measuring the average effect on a representative sample of the population is a flawed method of testing the efficacy of smartdrugs. These observations generalize to other chemical interventions designed to enhance normal human functioning.

Nick Bostrom
Dept. Philosophy, Logic & Scientific Method London School of Economics <>