Thursday 14 June 2012

Research Briefing: Can boosting motor inhibition help us resist temptation?


A lot of enjoyable things in life are risky and potentially addictive. So how do we control our impulses? And why do some people find it harder to say ‘no’ than others?

In a recent study we asked whether a key to self-control could lie in an unexpected place: a corner of our cognitive system that controls motor actions. We found that when people did a simple task that required starting and stopping finger movements, they also took less risk when gambling. This effect lasted at least two hours after being trained in so-called ‘motor inhibition’.

Why should the act of inhibiting simple movements lead to more cautious gambling behaviour? We don't yet know, but our working hypothesis is that it boosts or primes an inhibition system in the brain that regulates a range of functions - including complex decision-making. By strengthening motor inhibition through the mental equivalent of a ‘gym workout’ we may be able to open new avenues for treating problem gambling and other addictions.

-----------------------------------------------------------------------------------------------------------
Source article: Verbruggen, F., Adams, R., & Chambers, C.D. (2012). Proactive motor control reduces monetary risk taking in gambling. Psychological Science, 23, 805-815. [pdf] [press release]
-----------------------------------------------------------------------------------------------------------

Imagine the following scenario. You are driving to meet your financial adviser for a meeting about your investments. Along the way you encounter a series of obstacles that cause you to drive with extra caution: roadworks, speed cameras, and intermittent bursts of rain. When you eventually arrive and sit down with your adviser, she asks how you would like to spread your reserves between a number of low- and high-risk options. Choosing isn’t easy – the higher risk investments could pay for that much-needed vacation in the Maldives, but the market is unpredictable and you could lose out. You make your choices.

Clearly this decision is complex and based on many different sources of information. But ask yourself: would your decision have been the same if the journey to the meeting had been free of obstacles? Intuitively, you’re probably thinking “Huh? I would select my investments rationally, why should the drive there make any difference?” And most people would agree with you – society reinforces the notion that being able to make decisions rationally and without bias is part of what ‘makes us human’.

There’s just one problem with this argument: it doesn’t quite fit the evidence. Previous research tells us that multitasking impairs cognition, and we also know that priming people in various ways can bias social attitudes and financial decisions that we would intuitively ascribe to our free will. A recent study, for example, found that priming people with the mere image of a thinking man reduced their religious beliefs. At the same time, taxing self-control can cause what social psychologists call ego depletion, reducing our ability to resist temptation.

So, are you still sure that your cautious driving would have no effect on your investment decisions?


Spreading caution around


If our ability to make rational decisions can be influenced by cognitive interference, then you might assume that such effects should impair decision-making. Some evidence does indeed suggest that taxing executive control can make it harder for people to inhibit impulsive choices, although not all studies agree.

But what if we could specifically tailor a kind of multitasking that would improve your decision-making? In other words, what if the interference somehow biased you to take less risk, like the example above with cautious driving? To test this idea, we designed a laboratory task that brings together two different forms of decision-making: monetary gambling and basic stopping of a motor response.

Here’s how it worked. On each trial of the task, people were presented with six options below a series of yellow bars. Each of these options was a number of points that could be won, which – depending on the condition – ranged from 2 to 448. Higher amounts were intuitively more attractive but, crucially, also had a lower chance of winning. And if you did lose the gamble then you forfeited half the amount wagered. So, for instance, if you picked ‘112’ you had only a 15% chance of winning the 112 points, but an 85% chance of losing 66 points. Whereas if you picked ‘2’ you had a 75% chance of winning those 2 points, and only a 25% change of losing 1 point.

We didn’t tell people the exact probabilities of winning or losing, but we did tell them that the chances of winning were lower for higher amounts. We then calculated a simple betting score by taking the average of the choices across the options: from 1 to 6, ranked in order of lowest risk to highest risk. This means that the higher the score, the more willing the participant was to take risks when gambling. At the end of the experiment participants were paid the overall amount they won, at a rate of 1000 points to £1.

On each trial of this task, participants were given a few seconds to reach a decision before the yellow bars started rising toward a white line. Once the bars reached the line they then pressed whichever key corresponded to their choice. This was followed by feedback as to how many points they won or lost on that trial, plus a readout of their overall points balance.

Our behavioural task for combining monetary gambling with motor inhibition. The upper panel (A) shows a typical sequence of stimuli on trials without motor inhibition (called  ‘no-signal’ trials). The first screen (left) presented the various possible choices, ranging from smaller low-risk amounts, to great high-risk amounts. The letters below each option reminded participants which key on the keyboard corresponded to which choice. After 3.5 seconds the bars began to rise and participants made their response when the bars reached the white line. They then received feedback indicating whether their wager was successful or not, and their overall points balance. The lower panel (B) shows a sequence of stimuli on a trial that involved motor inhibition (‘signal’ trials). Everything is the same as (A), except that the bars now turn black just before reaching the white line.

To test how stopping of simple responses (i.e. motor inhibition) interacts with gambling decisions, we introduced an additional catch. Sometimes the bars would turn black just before reaching the white line. On these trials, participants were told to stop whatever decision they had planned. If they stopped successfully then they would win points, but if they failed to stop they would lose points.

The critical manipulation in this experiment was the expectation of stopping. To achieve this we further split the task into blocks of trials in which participants either expected ‘stop signals’ to occasionally occur (dual-task blocks, so named because these blocks included two tasks, gambling and stopping) or in which they were told in advance that signals would never occur (single-task blocks, so named because these blocks only included the gambling task).

We then compared the average betting scores between the blocks, focusing specifically on the trials without stop-signals. This allowed us to directly compare the effect on gambling of either expecting or not expecting to stop a response, while keeping everything else the same. In other words, the only thing that differed between these two conditions was the participant’s cognitive expectations.

So what did we predict would happen? There are two main possibilities. On the one hand, when people were in dual-task blocks they were now dividing their attention between two tasks. It is possible that this state of divided attention and cognitive ‘load’ could interfere with decision-making in the gambling task, making it harder for people to resist the more tempting, higher-risk options. We called this hypothesis the interference account.

On the other hand, we also know that when people expect to stop a response they become more cautious in their motor control – mainly, they slow down. So could this state of motor cautiousness transfer or spread to other forms of decision-making? If so, then when people expect to stop their response in the dual-task blocks, they might actually become more cautious and so take less risk than in the single-task blocks. We called this hypothesis the transfer account.

So which hypothesis won in the contest between interference and transfer? The results clearly supported the transfer account. When people expected to stop their motor response, their betting score decreased by 10-15% compared with when they knew they wouldn’t have to stop. So when people were expecting that they might have to stop their response, they freely chose to place less risky bets.

To be sure that this effect was specific to motor inhibition, rather than attention or other general effects of cognitive load, we also tested another group of participants in a ‘double-response’ control condition. Rather than stopping their response on signal trials, participants in the double-response group made an extra response. The double-response group showed no such reduction in risky gambling (in fact, it increased slightly), which helps tie the effects in the stop group to inhibition. And to be sure that these findings weren’t a statistical fluke, we ran the whole experiment twice in different participants to replicate the main finding.

The results of our first experiment. The left figure (A) plots the average betting score in the two groups of participants (double-response vs. stop) and for the two different conditions (single task vs. dual task). A higher betting score indicates riskier betting behaviour. Notice how the betting score is reduced under dual task vs. single task conditions in the stop group only (arrow; red bar). The right figure (B) shows the distribution of choices in the stop group, from the lowest risk (1) to the highest risk (6). Notice how expecting to stop a response  in the dual-task condition increased the proportion of lowest-risk responses (arrow) compared to blocks where participants never expected to stop (single task condition).

What do these results signify? From a theoretical perspective they reveal an overlap between different forms of inhibition: inhibition of motor responses causally shaped inhibition of risky gambling decisions. Previous studies have hinted at that such links might exist but much of this evidence relies on correlation rather than causation. For instance, people with a gambling addiction can sometimes show impairments in motor inhibition but it is unclear whether these problems are causally related.

Having uncovered evidence for a causal link we next asked whether training people in motor inhibition could have a more lasting effect. If so, this would suggest that the relationship between motor inhibition and risk-taking behaviour might be developed as a complementary therapy for addiction.


Bootcamp for inhibition?


In the next series of experiments we asked whether training people to stop responses could reduce risk-taking later in time. The idea was to train people for a short period (about 30 minutes) at motor inhibition, followed by monetary gambling. The gambling task was the same as described above but including the single-task blocks only, i.e. the bars never turned black and participants never expected to stop their responses while gambling.

We began by dividing people into three training groups. The stop group did a standard motor inhibition task, called the stop-signal task. The double-response group did a different (non-inhibition) task on the same stimuli. The control group didn’t do any training – they just skipped straight to the gambling task.

The stop-signal task is a workhorse of experimental psychology made famous by Gordon Logan, and one of the most straightforward and elegant tests of cognitive function. In our version of the task, participants were shown a shape on a computer screen (square or diamond) and were asked to identify the shape as quickly as possible by pressing one of two buttons, e.g. left button for a square vs. right button for a diamond.

On a random third of trials, the shape turned bold after a short delay. These trials are called ‘signal trials’ and the participant is instructed to try and stop their response. Successfully stopping your response is easy when the signal occurs immediately after the shape appears, but it becomes progressively more difficult as the delay between shape and the stop-signal is increased. This is because, at longer delays, you will be closer to executing your initial response by the time the signal occurs, so there is less time to countermand that response.

Our double-response group did a control task on the same stimuli: instead of trying to stop their response on signal trials, they instead executed a second response. So their task had similar attentional demands as the stop-signal task, but crucially without requiring motor inhibition.

The training phase in our second series of experiments. On ‘no-signal’ trials, participants decided as quickly as possible whether the stimulus was a square or a diamond. On a third of trials, the shape turned bold after a variable delay, termed a stimulus onset asynchrony (SOA). How participants responded on these ‘signal’ trials depended on which group they were in. Those in the stop group attempted to cancel their original response, while those in the double-response group made a second response. The numbers in the figure indicate the duration of the different events, in milliseconds.

So what might happen if we give participants the stop-signal task followed by the gambling task? If the effect of motor inhibition transfers over time to risk-taking behaviour then we expected training to make people more cautious in their gambling decisions, producing a similar effect to the first series of experiments. On the other hand, requiring people to continuously start and stop for 30 minutes might fatigue their inhibitory control and lead to increased risk-taking.

Once again the results were clear: motor inhibition training reduced risky gambling by 10-15%. Interestingly, we saw the same pattern even when we introduced a 2-hour gap between the end of the stop training and the start of the gambling task.

Training in motor inhibition reduced risk-taking in the gambling task by 10-15%. Note how the red bars are lowest when the gambling task immediately followed training (left set of bars), even after a two-hour delay was added between the training phase and the gambling phase (right set of bars).

A picture takes shape…


To summarise, we found that when people expected they might have to stop a motor response in a gambling task, they opted for less risky choices. And when we trained people to stop motor responses before doing the same gambling task, they also selected less risky options. This post-training aftereffect lasted for at least two hours. Overall then, these results indicate that these very different types of cognitive control are tightly coupled.  

Why such a link, you might ask. One possibility is that motor inhibition and risky decision-making draw on the same regulatory systems in the dorsolateral prefrontal cortex (DLPFC), a complex and mysterious part of the brain that coordinates a range of executive functions.

Of course, since these experiments are purely psychological, we can’t draw any conclusions about what might be changing in the DLPFC, but there are several possibilities to consider in future studies. For instance, recent work has found that more impulsive people tend to have lower levels of an inhibitory neurotransmitter called GABA in their DLPFC. Could motor cautiousness and inhibition training be somehow altering the expression of GABA in the DLPFC? Is motor cautiousness somehow tuning neural networks that regulate our behaviour, strengthening or biasing a computational ‘muscle’ that is used for decision-making? Perhaps inhibition training boosts the activity of DLPFC in regulating more primitive parts of the brain that respond to emotion and reward, such as the amygdala? Such questions are speculative, so to learn more we are now combining motor inhibition and gambling with a range of neuroscience methods, including transcranial magnetic stimulation (TMS), fMRI, simultaneous TMS-fMRI, and magnetic resonance spectroscopy.

As well as helping us understand more about cognitive control, our findings also have possible implications for treating gambling addiction. Related work by Katrijn Houben and Anita Jansen suggests that motor inhibition is linked to other compulsive behaviours, such as overeating and alcohol consumption. So could a regime of motor inhibition training help people overcome addiction? It seems possible, but we can’t claim from our results that motor inhibition provides a cure or treatment for any addiction. It is important to stress that all of the experiments in our study included healthy people only, and we currently have no data on whether motor inhibition training has any beneficial effect in a clinical situation. Furthermore, the effects we found are modest, just a 10-15% reduction in risk-taking. That said, we think the clinical angle is worth exploring and we may be able to tweak the design to make these effects larger and more clinically significant.

So can motor inhibition help us resist temptation? Possibly, yes. The next challenge is to figure out why and explore the implications – and applications – in clinical psychology and psychiatry.

____

* All comments and questions are welcome.

* Thanks to Frederick Verbruggen for comments on a previous draft of this post.

* The press release associated with this study follows a new format arising from the recent Royal Institution debate we took part in on science and the media, hosted by Alok Jha and Alice Bell, and also featuring Ed Yong, Fiona Fox, and Ananyo Bhattacharya.