Sarah Lazarus
Triplets stand around a roulette table. Each one has been there all evening, and each is preparing to hedge another bet.
Sibling 1: usually gambles sober. Except tonight, they are drinking.
Sibling 2: typically gambles intoxicated. Yet tonight, they are dry.
Sibling 3: always gambles drunk. Tonight is no exception.
Spoelder et al. says which triplet this evening will place the least advisable bet. Now Spoelder’s lab did not have numerous triplets to study, nor a large group of people with equivalent up-brings and genes. They had rats—lab bred rats with the same diets and housing and who could all be placed under tight experiment control. This allowed Spoelder et al. to study the direct effects of alcohol on decision making without the complication of human differences.
Researchers taught a group of rats how to gamble. There were levers for three options, each (‘safe’, ‘risky’, and ‘optimal’) with different chances of administering prize sugar. Rats trained unintoxicated quickly learned that ‘optimal option’ was their best bet. What happened when the researchers subsequently gave these rats alcohol? Those animals actually made only slightly different decisions. We can guess it is because they had already established their decision patterns while sober.
Yet what if you learn to gamble while drinking? A new group of rats were taught to gamble while intoxicated, then were tested sober, and then again while intoxicated. Rats trained while under the influence made far riskier decisions during their learning period. The reason, Spoelder et al. conclude, was because rats with alcohol in their system were bad at using feedback to adjust their choices. They had difficulty realizing the ‘optimal choice’ was optimal. Yet when the alcohol-trained rats were sober again, they made as many optimal decisions as their permanently sober counterparts; they were now using feedback to adjust their choices. Finally, these rats gambled once again under the influence of alcohol. Note that this time each rat now had the experience of gambling while sober, meaning they had at some point practiced ‘good’ decision making. It may explain why in this final trial rats made comparable decisions to their sober-trained counterparts. Further, this rationalizes why the poorest and riskiest choices were made by intoxicated rats with no sober experience.
Spoelder et al. tell us it is not Sibling 1 who will make the riskiest decision—for the person who learned to gamble sober will likely remember how to make optimal decisions. And it is not Sibling 2, who, now sober, has the chance to use feedback and find their best bet. It is Sibling 3. Someone who has only gambled drunk has not internalized various risks and is therefore least likely to choose wisely.
Humans are not rats and cages with levers are not casinos. Yet disentangling a person’s own tendencies with the direct effects of alcohol is difficult. Spoelder et al. are able to make the precise link that alcohol inhibits the ability to use feedback, and without prior experience, this leads to suboptimal decisions.
References: Spoelder M, Lesscher HM, Hesseling P, Baars AM, Ktooster LG, Mijnsbergen R, Banderschuren. Altered performance in a rat gambling task after acute and repeated alcohol exposure. Psychopharmacology