The Guess Worker

The obvious conclusion is not always the correct conclusion.

The sun rises in the East, travels in an arc across the sky and then sets in the West. Seeing as the Earth does not seem to move, the obvious conclusion is that the sun orbits the Earth. But this is, of course, the wrong conclusion.

The experiments

Kent Berridge and his fellow researchers have arrived at some conclusions about motivation through various experiments with dopamine.*1 Here is a much simplified summary of the results to these experiments and the conclusions drawn from them *2:

Experiment 1.

Dopamine deficient rats show very little motivation. They are listless and inactive and won't even feed unless forced.

Conclusion: Dopamine is necessary for motivation.

Experiment 2.

When these dopamine deficient rats are fed sugar solution, they exhibit facial expressions which indicate that they like the solution. They also exhibit facial expressions which indicate they dislike sour solutions.

Conclusion: Liking and disliking do not depend on dopamine.

Experiment 3.

Genetically modified mice which produce excessive dopamine do not show a greater liking of sugar solution compared to normal mice. If anything, they seem to like the solution less.

Conclusion: Liking does not depend on dopamine.

Experiment 4.

 These genetically modified mice show a greater wanting for sugar solution compared to normal mice.

Conclusion: Wanting depends on dopamine.

We can state these conclusions in one sentence: dopamine motivates through the feeling of wanting but has no role in liking and disliking.

The static versus the dynamic

Even though the scientists conducting these experiments may not have been aware of it, in making their conclusions they were also proposing a model for dopamine's mechanism of action. The model is a simple one, which is why they may not have realised they were proposing it. This model, which I'll call "the static model", assumes that fixed concentrations of dopamine are responsible for feelings.

These scientists used the static model to draw the obvious conclusions. If true, their conclusions could be very significant. Previously pain and pleasure were the only motivational feelings that we knew about. Now, though, apparently wanting has been shown to be independent of pain and pleasure. It seems that scientists have uncovered a third kind of motivational feeling.

I have argued that pain is the only feeling which motivates directly and that wanting is a type of pain. In my last post I proposed a model to explain how dopamine affects our feelings. In that model rises in dopamine concentrations cause feelings of pleasure and falls cause feelings of pain. Wanting therefore is caused by falling levels of dopamine. I'll call this "the dynamic model".

In the dynamic model, then, wanting is not independent of pain. Does this mean that the results to the experiments show that the dynamic model is wrong?

The experiments again

To answer this question we have see whether the dynamic model could give the same results. Here are the results again, but this time interpreted according to this model:

Experiment 1.

Dopamine deficient rats show very little motivation. They are listless and inactive and won't even feed unless forced.

Conclusion: the brains of rats with very low levels of dopamine can't easily create rises of dopamine when they perceive food stimuli. If there is no rise in dopamine there is also no subsequent fall. In consequence they can't feel an initial liking followed by a wanting. In other words the rats aren't attracted to food.*3 Therefore they aren't motivated to eat.

Experiment 2.

When these rats are fed sugar solution, they exhibit facial expressions which indicate that they like the solution. They also exhibit facial expressions which indicate they dislike sour solutions.

Conclusion: even though dopamine deficient rats have very low concentrations of dopamine they still have some (between 1 to 10% that of normal rats). Absolute levels of dopamine aren't important, only rises and falls. So as long as there is some dopamine, small falls and rises are possible. This means such rats can experience liking and disliking. These feelings are however not strong enough to motivate the rats to feed themselves.

Experiment 3.

Genetically modified mice which produce excessive dopamine do not show a greater liking of sugar solution compared to normal mice. If anything, they seem to like the solution less.

Conclusion: brain synapses which already have high concentrations of dopamine can't easily raise this concentration. This means that rises in dopamine are less likely and if they do occur the rises won't be as sharp as those in mice with lower levels. As a result mice with excess dopamine don't like sugar solution as much as normal mice.

Experiment 4.

These genetically modified mice show a greater wanting for sugar solution compared to normal mice.

Conclusion: brain synapses which already have high concentrations of dopamine can easily lower this concentration. This means that falls in dopamine are more likely and when they occur the falls will be sharper than those in mice with lower levels. As a result, mice with excess dopamine want sugar solution more than normal mice.

Which model do you prefer?

The dynamic model, then, can explain the results as well as the static model can. Which of the two models is more likely to be correct? The best way to decide whether a model is correct is, of course, to see if all experimental results correspond to the model's predictions. The problem is that different animals under the same conditions often behave in different ways. This means that experimental results are sometimes contradictory and difficult to interpret.

Even so, in general, results to a range of experiments do seem to fit the dynamic model better. In experiment 3 above, for example, mice with excess dopamine appear to like sugar solution less than normal mice. In the static model there is no easy explanation for this. Liking should be independent of dopamine and so higher dopamine levels should have no effect on liking. But apparently there is some connection between the two.

Beyond formal experiments, the dynamic model seems to provide more elegant explanations for the way our feelings work. The static model, in contrast, raises a host of questions which can't easily be answered.

In the dynamic model, wants motivate because they are a kind of pain. It is easy to understand how pain motivates us. But if wanting is independent of pain, as in the static model, how does it motivate us? By what mechanism does wanting make us do things, if not through pain?

Through our own experiences we know that for every pain we feel, we also feel a want. If I have a throbbing headache, I want to soothe it. If we feel a pain, at the same time we also feel a want to get rid of the pain. Pain and wanting are therefore connected. This connection makes sense in the dynamic model. But it is not easy to understand this connection if wanting is supposed to be a feeling independent of pain, as in the static model.

And what about pleasure? We usually want things that we like. If we can't get the things we like, our wants often feel painful. When we satisfy a want we feel pleasure, and when we satisfy a painful want we feel even greater pleasure. Why do such connections between the three feelings happen? Again, these connections are explained in the dynamic model but the static model leaves the question unanswered.

The obvious is more complex

The geocentric model - that is, the idea that the sun orbits the earth - was the obvious conclusion at one time. But as more information about the movements of the stars and the planets across the sky was gathered, the geocentric model became less and less tenable. To make the geocentric model work, the orbits of some celestial bodies would have to be almost impossibly complex. The heliocentric model - with the sun at the centre of the solar system - made the orbits simpler: everything suddenly fitted into place.

The static model provides the obvious conclusions as to how dopamine affects feelings in the brain. To make the static model work, though, the answers to the questions it raises would have to be almost impossibly complex. With the dynamic model, on the other hand, everything seems to fit into place. So even though the dynamic model is less obvious, it provides a simpler and neater explanation to what we observe. This is why it is more likely of the two models to be correct.

There are another couple of pluses for the dynamic model. It gives an indication of how feelings are regulated and why they fluctuate in intensity. There'll be more on this in future posts.

*1. For more information about Kent Berridge and his ideas see my posts "A conversation with Kent Berridge" and "Dopamine".

*2. For more details on these experiments see the following papers:

i) Smith, Berridge, Aldridge: Disentangling pleasure from incentive salience.., 2011, PNAS

ii) S.Pecina et al.: Hyperdomaminergic Mutant mice have higher "wanting" but not "liking" for sweet rewards, 2003, Journal of Neuroscience, 23, p. 9400

iii) Roy A.Wise: Dopamine, learning and motivation, 2004, Nature Reviews, Neuroscience, Vol.5

iv) K.Berridge, T.Robinson: What is the role of dopamine in reward..?, 1998, Brain Research Reviews 28 (309-369)

*3. See  Attraction