The Guess Worker

A conversation with Richard Palmiter

  • Richard Palmiter is Professor of Biochemistry at the University of Washington, Seattle.
  • He and his group at the University have been examining the role of dopamine in motivation, reward and learning.
  • I wrote to him about his research and this led to the following exchange of emails.

1) 17.6.2011

Dear Professor Palmiter,

I am fascinated by your research on the effect of dopamine on motivation in mice. I would like to trouble you by suggesting a model for the mechanism of dopamine motivation. I hope that you will find the following in some way useful.

In normal mice, after (let's say) a visual stimulus of food, there will be a spike of dopamine in some synapses of the brain. This spike occurs because of a previously learnt association between the stimulus and pleasure. The initial rise in dopamine concentration is followed by a fall - probably caused by uptake.

I believe that the rise in dopamine level is experienced as a "liking" and the subsequent fall is experienced as a "wanting" for the food. It is the wanting phase of the spike, not the liking phase, which provides motivation.

Whereas increasing concentrations of dopamine are interpreted by the consciousness as pleasure, decreasing concentrations are interpreted as pain. (By "pain" I mean the whole range of unpleasant sensations from the most intense physical pain to barely perceived discomfort).

Pleasure cannot motivate - at least not directly. Pain, on the other hand, is the only direct motivator because it drives animals into taking actions to remove it.

Under most circumstances, wants are minimally painful and may not even be recognised as being unpleasant. Nevertheless, even minimal pain is sufficient to motivate. A transiently registered want may be enough to motivate normal mice to feed.

Dopamine depleted mice, though, can have no dopamine spike. Therefore they cannot experience wants. Hence their lack of motivation.

I would, of course, greatly welcome any feedback from you on the above ideas.

Kind regards, Philip Ludikar

Praha, Czech Republic

2) 19.6.2011


One problem with your intriguing view of dopamine action is that aversive events can also lead to a spike in dopamine neuron activity from some dopamine neurons (see our recent Zweifel et al paper in Nat. Neurosci - ).  We think that the dopamine neurons that spikes  provide an alerting signal (pay attention to that sensory stimulus) and those that are inhibited provide a valence signal (something good or bad happened).  When we dampen the alerting signal but leave to valence signal intact, mice develop a generalized anxiety - i.e. they do not associate the bad with anything specific (because they weren't paying attention).  Where these different signals are detected in the brain and which receptors are involved are major questions.

Another thing to keep in mind is that spikes in dopamine release preferentially activate D1 receptors, and they are located on different cells than those that are activated by D2 receptors. Whereas decreases in dopamine will preferentially affect those cells that express D2 receptors.

I don't think "liking" has much to do with dopamine since mice without any dopamine still like sucrose (i.e. given a choice they prefer sugar solution over water).


3) 19.6.2011


Thank you indeed for your reply.

I don't think that a spike in dopamine neurone activity in adversive events necessarily contradicts my model. This would depend on what is happening in the synapses during such activity. If there is a sharp fall in dopamine concentration following a rise, then the fall would be interpreted as an unpleasant sensation (ie. pain).

I can also see no contradiction in the idea of alerting and valence signals. I believe that consciousness is unifocal, which means that the mice can only be aware of one kind of information at a time - whether this information is pain, pleasure or an external stimulus. Attention and concentration are words used to describe the amount of time consciousness spends on each set of information. In general, at any moment, animals are going to be aware only of the most pleasurable or painful sets of information. This means that consciousness will always be centred on the parts of the brain with the most dopamine activity.

It seems to me that if you dampen alerting signals, you are blocking the pathway between external information (tail pinching) and the consciousness. Little or no time can then be spent by the consciousness on this stimulus and so the mice don't "pay attention". Therefore no concepts (ie. associations) can be formed between the specific information of tail pinching and the unpleasant sensation, which is conveyed by the valence signals.

As you imply, the failure to make such an association is the reason for the resulting anxiety. Anxiety, being painful, is intended to be motivational. However, in this case, efforts to find an association cannot be successful. This results in a useless increase in the intensity of the pain.

Regarding "liking"- my use of the word is different to yours. By "liking" I mean an attraction or pleasure felt towards a visual stimulus, such as experienced upon seeing food. Here "liking" plays an (indirect) motivational role as it causes the subsequent "painful" want.

Best regards, Philip Ludikar

4) 19.6.2011


Most drugs of abuse and direct-acting dopamine receptor agonists give a prolonged occupancy of receptors (min to hour) followed by a slow decline (due to metabolism).  Animals and people respond behaviorally during the prolonged occupancy stage, and will perform actions to maintain the occupancy of dopamine receptors. I suppose one can argue that people continue to take the drugs because they don't like the unpleasant effects of drug withdrawal, rather than liking the effects of the drug itself, or both.

For every spike in dopamine signaling, there has to be an up-phase and a down-phase - so it is not easy to know whether the increase or the following decrease in read as a signal.

In your view, would rapid lowering of dopamine neuron signaling from baseline also be interpreted as an unpleasant "wanting" signal - i.e. motivate an animal to perform actions that would repeat the lowering of dopamine neuron activity?  Is the fall from a previous rise that is important for wanting, or is it just a fall from baseline.  What about an animal that has high basal dopamine signaling all the time (e.g. mice that lack a dopamine transporter) - would they "want" to perform actions that would lower their dopamine?

Techniques are now available (e.g. using channel and halorhodopsin) to provide transient spikes or depressions in dopamine signaling independently of other events. The results so far, indicate that mice quickly learn to perform actions that promote spikes of dopamine neuron activity, but I haven't seen anything published yet for transient depressions in activity.


5) 21.6.2011


I'll answer your email point by point.

"I suppose one can argue that people continue to take the drugs because they don't like the unpleasant effects of drug withdrawal, rather than liking the effects of the drug itself, or both."

As you know, my argument is that people continue to take the drugs only because they don't like the unpleasant effects of drug withdrawal and not because they like the effects of the drug itself.

There are fundamental difficulties with the idea of pleasure being an instrument of motivation. Let's say an animal is already getting pleasure from some activity. It is hard to see why that pleasure would make the animal seek more of the same pleasure. If it's already got pleasure, why would it need more of it? And assuming that pleasure could motivate, when would the endpoint come? When would the pleasure be enough so that the animal would no longer seek more? And why would an endpoint come at all?

In contrast, the way in which pain motivates is simple. An animal takes actions to get rid of pain. It will continue taking actions until the pain is completely removed. The endpoint is therefore clear.

The problem with the idea that only pain motivates, of course, is that our own experiences seem to tell us that pleasure often motivates us. I would say these are illusions caused by two main factors. First, our wants may be so minimally painful and acted upon so quickly that we may barely perceive them. Second, our memories contain many concepts associated with pleasure. These concepts tell us how to achieve pleasure in future situations. But they may be used unconsciously, which means no motivation is involved.

In my view, then, all motivated actions must be caused by pain. Seeing as your research shows that dopamine is involved in motivation, this suggests that there could be a link between dopamine and pain.

"For every spike in dopamine signaling, there has to be an up-phase and a down-phase - so it is not easy to know whether the increase or the following decrease is read as a signal."

I would guess sometimes both, sometimes only the up-phase, sometimes only the down-phase and sometimes neither. As I wrote earlier, I believe consciousness is unifocal, which means that animals are only going to be aware of one set of information at a time. In general this will be the most important information. The most important information will be signalled by the parts of the brain where there is the most dopamine activity. Specifically this will be where there are the steepest rises and falls in dopamine concentration.

"In your view, would rapid lowering of dopamine neuron signaling from baseline also be interpreted as an unpleasant "wanting" signal - i.e. motivate an animal to perform actions that would repeat the lowering of dopamine neuron activity? Is the fall from a previous rise that is important for wanting, or is it just a fall from baseline."

I think that potentially any fall could be interpreted as painful and any rise as pleasurable. Whether the pain is experienced as a want would depend on associations formed.

I believe the baseline is important in the regulatory mechanism. Rises above the baseline will be conteracted by a tendency for the dopamine level to fall back to the baseline. Conversely, falls below the baseline will tend to be followed by a rise to the baseline - unless the brain continues to receive information which keeps the dopamine level depressed. The experiences of "satisfaction" and "relief" could occur because of a rise in dopamine level back to the baseline.

"What about an animal that has high basal dopamine signaling all the time (e.g. mice that lack a dopamine transporter) - would they "want" to perform actions that would lower their dopamine?"

I don't think they would "want" to perform actions that would lower their dopamine. Just as in normal mice, a want could only occur if dopamine should happen to fall. But the behaviour of high basal dopamine mice may be affected by "equilibrium pressures". Possibly in these mice there could be a greater "pressure" for dopamine levels to fall back to the baseline. Sharper falls would mean more intense pain and therefore more keenly felt wants.

Philip *

6) 1.4.2013

Dear Richard,

I wrote to you a couple of years ago to suggest a mechanism for dopamine's action in motivation. (See emails below.) Since then I have found the time to study thoroughly your 2011 paper "Activation of dopamine neurons is critical for aversive conditioning". It seems to me that your experimental results are entirely consistent with the model I proposed. Where we differ is in some of our interpretations of these results.

You describe the activated dopamine neurons during aversive stimuli as providing "motivational salience signals". I would suggest that the notion of incentive salience is superfluous. All motivated behaviours can be explained as being driven by pain and pleasure. Experiments which appear to show that liking and disliking are independent of wanting can be successfully explained in an alternative way (see "Is Kent Berridge wrong?" on my website http://philipludikar.com). Furthermore, unlike with pain and pleasure, no convincing mechanism has been proposed to show how incentive salience is supposed to motivate.

I agree with you that dopamine neurons which are inhibited with aversive stimuli are likely to be those signalling motivational value. I believe what happens is that, in specific synapses of these neurons, there are sudden falls in dopamine levels, which are interpreted by the brain as unpleasant or painful sensations. Painful sensations motivate the brain to search for information which may help to alleviate the pain. In most cases (and possibly in all cases) the brain is searching for information which can be recognised: that is, for information which corresponds to associations already stored in the brain.

When the search reaches a successful outcome, a new association between valence and salience neurons is made. This association is, of course, quite likely to be a synaptic connection.

In your paper you refer to other research which showed that dopamine is released into the amygdala during stress. It is possible, as you suggest, that dopamine is important for stimulus processing. That would mean that dopamine plays a role in the formation the synaptic connection mentioned above.

However, there is another possibility. Dopamine release could be a consequence (rather than the cause) of successfully formed synaptic connections. This could explain why, let's say, we feel satisfaction upon solving a difficult crossword puzzle clue. Here the unpleasant sensation is minimal (although perhaps felt as frustration) and the subsequent rise in dopamine (after making the connection) is experienced as pleasure.

It is unlikely that footshocked mice experience pleasure when they form synaptic connections. But according to your results their anxiety is lower than in knockout mice. In other words, repeated footshocks are "less unpleasant" for control mice. I would argue, therefore, that forming a connection between valence and salience neurons must result in the moderation of the sharpness of dopamine falls.

As I appear to be suggesting that only valence neurons are responsible for motivation, it has to be asked why salience neurons are also activated by averse stimuli. I would guess that this is because all sensory stimuli have the potential to elicit some kind of pain/pleasure sensation. This sensation could be, for example "interest". In the presence of aversive stimuli such as footshock, it is unlikely that these sensations would be experienced. Attention would be diverted to the stronger signals for pain.

This would mean that salience neurons can also be responsible for motivation in certain situations. However, this motivation would still operate through pain and pleasure.

These proposed mechanisms may or may not be correct. However, as long as pain and pleasure alone can be shown to do the job, there is no need to introduce a mysterious third driver of motivation. And if pain and pleaure can truly be proved to be inadequate, it is still simply not good enough to invoke incentive salience. A convincing hypothesis for its mechanism of action must be put forward. Otherwise, in my opinion, the concept of incentive salience is without value.

I would be happy to hear your thoughts on the points I have made.

Regards, Philip Ludikar *

* Because Richard Palmiter did not reply to these last two emails, I can only presume that he didn't feel a need either to defend his ideas or to argue why he thinks I am wrong. In this respect he differs from Kent Berridge, who in his courteous emails was always ready to debate the points I made.

Comments powered by CComment