Effects of Alcohol

Many drugs can slow responses in the tests we use, but alcohol is rather different. Most tests that we use have two measures: one of speed (e.g. reaction time) and one of accuracy (e.g. number of correct and incorrect responses). If you compare alcohol and temazepam using a test such as the four-choice reaction time, and look at both reaction time and errors, we see:

Read on to find out more about:

Speed, Accuracy and Alcohol

The Speed-Accuracy Trade-off and Risk

What makes Alcohol Special

Speed, Accuracy and Alcohol

Comparing alcohol and temazepam on the Four-Choice Reaction Test gives a different pattern for the two drugs:

Two doses of alcohol (Alc1 and Alc2, the higher taking volunteers to about the UK legal limit of 80 mg/100 ml) and two doses of temazepam (T, 15 and 20 mg) were compared to placebo (Pla). Compared to placebo, both alcohol and temazepam show a dose-related increase in mean response time (light bars) the increase being about 7% for alcohol, 10% for temazepam, compared to placebo. But alcohol also caused a large increase in errors (dark bars) -  errors were more than doubled at the higher dose compared to placebo. Temazepam showed no significant increase in errors (Tiplady et al.,. 2003)

 Another way to look at the same data is to plot speed directly against accuracy:

This graph shows exactly the same data as the barchart above. Here, each point on the graph shows a treatment condition. The open circle is placebo, the triangles temazepam, the squares alcohol. A light grey fill shows the low dose of each drug, black the high dose. Reaction times for each condition are shown on the horizontal x axis, errors on the vertical y axis. So slowing will cause a shift to the right, and an increase in errors a shift upwards. This type of graph demonstrates the difference between the actions of the two drugs. The greater the  angle between the lines for the two drugs, the greater their difference in speed-accuracy profile.  

In the four-choice test, both drugs cause slowing, while only alcohol significantly increasing errors. Other tests show an even greater difference. For example, here are results for the Gibson Spiral Maze:

In this case, the alcohol and temazepam lines are nearly at right angles, indicating an almost complete dissociation: alcohol increases errors with no effect on speed; temazepam slows responses (increases reaction times) with no effect on errors.


The Speed-Accuracy Trade-off and Risk

If you are impaired by a drug, the sensible thing to do is to take things slower, giving yourself time to respond correctly. This is an aspect of the speed-accuracy trade-off. In many activities you can adjust performance between two extremes - very fast behaviours with lots of mistakes, or slower behaviour with few mistakes. The appropriate balance between speed and accuracy depends on the task. So if you are impaired by a drug, it may be inevitable that there will be impairment to either speed or accuracy (or both), but it may not be obvious which. Maybe as well as causing impairments, the drugs themselves have an influence on the speed-accuracy trade-off? Looked at this way the difference between alcohol and temazepam can be seen as a relative shift in the trade-off, with alcohol leading to faster, more error prone behaviour than temazepam.

In many situations, including driving, faster means riskier. And there is some evidence from driving simulators that people tend to drive faster on alcohol than on placebo (though no one seems to have compared this with other drugs). Maybe alcohol increases our willingness to take risks? Several research groups have looked at this, using situations in which volunteers take real, though not physical risks - usually in some form of gambling situation where they are risking some of their participation fee. These studies have not found that alcohol increases "risk-loving" behaviour. 

Another possibility is that alcohol impairs our perception and processing of errors. When we adjust our speed-accuracy trade-off, we have to be aware of the errors we are making. If we don't notice errors, then we can't take them into account in adjusting our performance, and so would tend to go faster. In one of our recent studies, volunteers reported errors on the Four-Choice Reaction Test as they performed the test. They made more errors on alcohol as expected, but the proportion of errors they noticed was the same (Acons et al., 2004). So we .do notice the errors. There is some evidence that error processing is a distinct process from error detection, and this may be impaired by alcohol.

What makes Alcohol Special?

While lots of research has been done on alcohol, surprisingly little has compared it directly to other drugs. In some respects alcohol appears to be similar to other impairing drugs - for example alcohol and temazepam have similar effects on memory. Many drugs have the potential to impair driving, and have been shown to be linked to traffic accidents. 

The evidence on drugs and driving is interesting. Laboratory studies have shown impairments on driving-related tasks with many drugs, some of which are very widely used. However it has been quite difficult to show the links between, for example, benzodiazepines and traffic risk. This link has now been established, at least for daytime use of benzodiazepines (Barbone et al., 1998), but the increased risks are modest. By contrast, alcohol has "jumped out" of just about  every study that has been done since the pioneering work in the USA in the 1960s (see right: BAC = Blood Alcohol Concentration))

It is tempting to speculate that the relatively fast, error-prone behaviour we see in lab tests with alcohol has something to do with driving risk. Perhaps when you are impaired by a benzodiazepine, you feel impaired, and slow down, thus partially compensating for the risk due to the impairment. With alcohol, you think you are doing fine, don't slow down, or perhaps even go a little faster - but you're just as impaired, and so the risk is greater. There is no direct evidence for this idea, but it suggests an approach for new research.

Data from Kruger et al. (1995)

One area we are currently pursuing is obtaining people's own accounts of what is going on. If you are aware that you are making more errors with alcohol, why do you accept the increased error rate? Why not slow down, as you do with other drugs? One thing is sure: there is still a lot of interesting new work to do on this old drug.