Wittgenstein and Freud

At the start of his career, Freud hoped to develop a science of the mind that would explain psychological phenomena in terms of physiological processes taking place in the brain. But at that time the tools and techniques necessary to explore the functioning of the brain simply did not exist. So, Freud had to rely on speculation, effectively trying to reverse engineer an account of the nervous system based on what people said about their thoughts and feelings and on how they behaved. In his “Project for a Scientific Psychology” (1895), he developed an impressive theoretical framework, but, after several months of frenzied work, he abandoned the project, never to return to it. Astonishingly, South African psychoanalyst and neuropsychologist Mark Solms (and others) claim that neuroscience is providing evidence to support Freud’s central claims. Can this be right?

Mark Solms’ paper “The Scientific Standing of Psychoanalysis” (2018) provides an excellent introduction to his ideas. It sets out the three core claims of psychoanalysis that he believes neuroscience provides evidence for. It explains how the clinical methods of psychoanalysts fit with the current scientific understanding of the brain; and it argues that psychoanalytic therapy achieves outcomes that are at least as good as other treatments in psychiatry. I will concentrate on the first two sections of the paper since these relate to Solm’s claim that neuroscience is proving Freud right. The debate about the comparative effectiveness of psychoanalysis is a separate issue.

Solms claims that psychoanalysis rests on three claims about the “emotional mind”, each of which, he claims, is testable and falsifiable:

1. Like other species, we are born with a set of innate needs. 
2. The main task of mental development is to learn how to meet these needs in the world.
3. Most of our methods of meeting our emotional needs are executed unconsciously and if they need to be changed, this can only be done by our becoming aware of them.

The first two points seem beyond challenge, which might seem a good thing, but it suggests that they are not really empirical claims. Can we imagine a living entity that has no innate needs? Not really, particularly given the way that Solms uses the concept of need. If we imagine an angel who in the traditional sense might be said to have no needs (i.e. no need to eat, no need to reproduce, no need to keep warm etc), we still need to imagine the angel doing something if we are to see it as alive. But if it spends all its time praising and serving God, then Solms will attribute to it a need to praise and serve God and a drive to do so, which generates pleasure when the drive is satisfied and unpleasure when it is not. So, the first core claim would seem to apply to any conceivable living entity, while the second core claim would apply to any living entity that was not perfect at satisfying its needs from the first moment of its existence. They are not empirical claims. Neither is ever going to be disproved or refuted. They provide a framework for looking at the activity of any living entity.

What about Solms’ third core claim? This claim captures the idea that evolution pre-programmes us to meet our needs but that the programmes sometimes need to be adjusted, and this requires consciousness (indeed, Solms would see this as the evolutionary explanation for consciousness). It may be true that the brain-as-predictive-computer analogy dominates neuroscience, but it is misleading to suggest that Freud thought about the mind/brain in these terms – even in the Project. This is not surprising since (electronic) computers did not exist when Freud was alive. Rather than claiming that our brains are prediction machines, he sketched out a theory of the brain where mental processes are explained in terms of the movement of quantities of physical energy through networks of neurons. So, even if the brain-as-computer analogy generates a framework that has striking parallels to the one set out in the Project, the parallels require a major reinterpretation of Freud’s claims.

Let’s examine Solms’ three core claims in a bit more detail, starting with the first one. Solms suggests that the needs we are born with are met autonomically up to a point, but beyond that point our needs will only be met if they become conscious or “make demands upon the mind to perform work”. He suggests that this is an empirical claim, but it is not. He is fleshing out a particular way of thinking about human beings and linking it to statements made by Freud. Solms (and colleagues) argue that as well as bodily “needs” (our need to eat, our “need” for the levels of oxygen and carbon dioxide in our blood to stay within certain bounds, etc.) we also have “emotional needs”. But the two types of need seem very different, and it is hard to fit them into one pattern. How is our lust drive (our need to find sexual partners) met up to a certain point autonomically? Similarly, surely our seeking drive (our need to engage with the world) operates entirely through consciousness? Solms can, of course, meet these challenges. Since he is creating a model rather than making empirical claims, he has a free hand. He can stipulate that certain types of need are met in one way, others in another. We may wonder how helpful it is to make very different things sound the same, but if someone wants to call apples and oranges, “roundfruit1” and “roundfruit2”, that’s up to them.

Solms claims that affects are important in psychoanalysis because they broadcast a need and thereby “release reflective or instinctual behaviours, which are hard-wired predictions (action plans) that we execute in order to meet our needs, e.g. we cry, search, freeze, flee, attack”. This seems a rather reductive account of the role that feelings play in human life (and in psychoanalysis), but for Solms the function of our emotions is to trigger behaviour in us, they do not have any significance beyond this. The approach encourages us to look at emotions in a particular way. If we cry, this is a programme-driven strategy to meet our need for something (and a correct or incorrect prediction that this behaviour will lead someone to do something that will satisfy our need). It is not something we just do or something we do even if we do not believe it is going to improve things. No, we (or rather our brains) are programmed to survive and anything we do must be seen in the light of this goal (or “intention”).

So, how many emotional needs or drives do we have? Freud was a dualist. His initial account had two drives (the self-preservative drive and the libinal drive). His later account also had two drives (the libidinal drive (into which the self-preservative drive had been subsumed) and the death drive). Since Solms does not understand that what is at stake here is the development of frameworks, he is surprised at Freud’s preference for dualistic accounts. He acknowledges that universal agreement has not been reached on the number of innate human needs (or as he puts it “needs in the human brain”), but suggests that they are likely to include the seven emotional needs that the US neuroscientist Jaak Panksepp included in his taxonomy: SEEKING (our need/drive to engage with the world), FEAR (our need/drive to escape dangerous situations), RAGE (our need/drive to get rid of frustrating objects) , LUST (our need/drive to find sexual partners), CARE (our need/drive to care for and nurture others) , PANIC/GRIEF (our need/drive to attach to caregivers), and PLAY (our need/drive to test out and establish our role in the social hierarchy). 

Solms’ (and Panksepp’s) list seems a very miscellaneous list, and one can easily imagine lots of other possible classificatory systems, e.g. seeking (looking for options) could be seen as something that plays a role in our efforts to satisfy drives rather than as a drive itself. But if you are building a model or a framework, you get to decide on the parts you want to use and how you want to put them together. If you want to use different parts from Freud and don’t have a strong preference for dichotomies, that’s fine.

Freud claimed that the drives and, in particular, the libido are quantifiable, but he never made any progress in quantifying them. Solms also claims the drives in his model are in principle quantifiable. He argues that the gap between the predicted sensory state of a self-organising system and its actual states is in principle measurable and since it is this gap that generates the demand on the mind for work, the latter must also be measurable. As with Freud, however, actual drive measurement remains an aspiration. Like the fact that the number of innate human needs has not yet been definitely determined, this reinforces the sense that what is going on here is not what it seems. It is interesting to note, for example, that if you managed to persuade Solms that some aspects of human life did not fit into Panksepp’s seven-need framework, the obvious solution would be to add an eighth need.

Solms suggests that we learn in order to establish optimal predictions as to how we can satisfy our drives in a given environment. Again, this is one way of thinking about things, but it is not an empirical claim. It also does not seem to me to be a way of looking at things that will resonate strongly with many people or indeed with many psychoanalysts. He suggests there are two reasons we need to learn, although he does not separate them clearly. One reason is that while evolution programmes us for dealing with certain types of dangers, we may encounter new types of dangers (for example, he mentions electric sockets). Similarly, there will be new types of context that we need to learn about – the appropriate response to a lion safely behind bars in a zoo is different to one at large in the bush. 

But learning is not just about new dangers and new contexts. It is also about handling conflicts between our drives and reality and between our drives themselves. This is standard psychoanalytic territory, but it does not fit easily into Solms’ model. Our emotional needs are supposedly just the brain’s built-in strategies for the preservation and reproduction of the organism of which the brain is a part, so why should they conflict with reality? Why does our lust drive impel us to limitless reproduction rather than as much reproduction as the context allows? Similarly, why do our drives conflict? Why hasn’t evolution done a much better job in making our drives compatible? In Freud, the picture of the ego struggling with the instincts that we as animals have is a natural and powerful one, but with the Solms model it is a bit disappointing that our biological computers don’t function better (and don’t seem to be improving).

Perhaps one shouldn’t make comments of this kind, since the better our brains worked, the less need there would be for consciousness. Fortunately, our programmes do need fine-tuning and so our needs have to become conscious. It may seem to us that this is about working out what matters to us and how we can give our lives meaning. But, in fact, it’s about enabling our brains to get better at making predictions in order to maximise attainment in terms of our (fixed) biological goals (of self-preservation and reproduction). The reprogramming process uses our feelings to impel us to improve our strategies. They are essentially signals. Negative emotions are prompts for the development of new strategies, positive emotions are signs that our strategies and/or our improvements to them are working. This is a rather unflattering picture of ourselves, and Freud certainly sought to puncture our rose-tinted view of ourselves, but typically his argument was that we are less rational than we want to think we are, not that we are machines.

So how does Solms explain the unconscious? As noted earlier, he claims that most of our needs are met autonomically, but he recognises that the fact that we are not aware of the processes that maintain the functioning of our circulation system does not relate to the Freudian unconscious. So, he offers a different explanation. He argues that a child cannot always solve the problems it faces and illegitimately (or prematurely) automatises some of the strategies it chooses. He calls the strategies that shouldn’t have been automated “the repressed”. They are not repressed because the ego does not like them; rather they are strategies the ego liked too much and automated when it should not have done so. In any event, they need to be brought back into consciousness to be modified (“reconsolidated”).

Solms refers to these prematurely automated strategies as “wishes”, which hints at Freud’s claim that a baby’s first response to unmet needs is to hallucinate their satisfaction, but what he really means is any automated strategy that does not work. For him, this means generates “prediction errors”, where the predictions are that the strategy will be implementable and will deliver drive satisfaction. The child’s strategy of trying to sleep with its mother will not (generally) fit with reality, and its strategy of getting rid of its father will – even if successful – not (generally) lead to the satisfaction of all of its drives (and maybe to the satisfaction of none). Such strategies need to be brought back to consciousness and modified. As we shall see shortly, this is where therapy comes in.

Solms suggests that the difference between psychoanalytic therapy and psycho-pharmacological methods of treatment is that feelings (symptoms) are seen as meaning something. This is true, but when Solms makes this claim, it is not quite what it seems. He is not claiming that feelings have meaning in the sense that words have meaning; rather they mean something in the sense that a flashing red light (generally) means something. The meaning of our (negative) feelings is that one (or more) of our drives is not being met. For Solms, emotional disorders (negative feelings that refuse to go away) reflect faulty strategies. They arise from unsuccessful attempts to satisfy needs and as such involve intentionality. But whose intentionality? The only possible answer is our brain’s! In Solms’ model, there is no room for what we normally think of as agency. The ultimate driver of everything is brain activity evolutionarily tailored to specific biological outcomes. Beyond that, there is just the adoption and modification of strategies achieved via negative and positive feedback (in the form of feelings).

I suspect that few psychoanalysts would spontaneously describe their work as “helping people change deeply automatised predictions that they cannot be reconsolidated in working memory”. But this is a flexible formulation and most things that psychoanalysts want to say could probably be translated into this vocabulary. But psychoanalysis is based on values as well as claims, and these are very hard to fit into Solms’ account, which is focused on the biological unit, the individual, and the satisfaction of their drives. In this sort of account, other people can only appear as a means rather than as an end. There is no scope for the idea of respecting another person as someone with their own thoughts, feelings and desires.

Similarly, on Solms’ account the issue for therapy is how to optimise the individual’s feelings, which are the expression of the demands made on their mind by their biological computer. Does this really capture what psychoanalysis is about? What if there are a number of solutions that seem likely to deliver similar levels of outcomes in terms of feelings – should the analyst point this out to the patient and let them choose? What if the patient happens to be in a life situation where a well-managed narcissistic strategy has a good chance of delivering lots of positive feelings with little downside risk – is this the strategy the analyst should help them automise? In Solms’ account, there seems little scope for the idea of helping the individual face the truth about themselves or become a more mature person, someone who has more genuine relationships with others and a more integrated relationship with themselves.

So what does Solms achieve? I think he does an impressive job in drawing links between what Freud said (or some of what he said) and the current model that neuroscience uses to understand the brain. But he does not show that neuroscience is proving Freud right. At best, he shows that Freud’s approach is compatible with the model neuroscience currently uses. In theory, the links Solms draws might make some psychoanalytic claims testable, but the links he makes are so high level that it is hard to know whether this will ever be possible (and whether psychoanalysts would recognise any future testable claims as accurate translations of claims they want to make).

Solms suggests that the core claims he outlines (and more specific claims he makes in explaining them) are empirical claims, but this is not the case. They describe a way of thinking about the human mind/brain. The issue is not whether this model is accurate but whether it is helpful – and helpful for what. Apparently, it is the model that neuroscientists currently find most helpful in their efforts to explore the functioning of the brain. One day this may change, or it may not. Psychoanalysts may find it reassuring that their approach is compatible with this model, but neither the model nor the detailed claims made in using it provide empirical evidence to support the claims that psychoanalysts make. Solms has demonstrated that it is possible to map psychoanalysis onto the neuroscientific model, but in the process what is most interesting about psychoanalysis as an approach to understanding human behaviour and as a form of therapy becomes hard to recognise or gets squeezed out altogether.

In a more recent paper, “Revision of Drive Theory” (2021), Solms develops his position in a number of interesting ways. Firstly, he makes it clearer that his approach diverges quite a lot from Freud’s. Some of these differences do not matter psychoanalytically. For example, in “Beyond the Pleasure Principle” Freud suggests that the perception/consciousness system is located in the cerebral cortex, whereas Solms claims that it arises from the core of the brainstem. Clearly this issue has no impact on psychoanalytic theory or practice. The other revisions Solms proposes do seem more significant. For example, he claims that the (repressed) unconscious and the id are very different things (and located in different parts of the brain) and that not all drives are bodily drives. In fact, Solms thinks revisions are needed in relation to: the fundamental mechanism of drive; the bodily sources of drives; the classification of drives; the relationship between drive and pleasure-unpleasure; and the relationship between drives and conscious qualities in general. Quite a shopping list!

I do not think it is surprising that in taking forward his mapping exercise Solms has encountered a significant number of areas where the mapping does not work well and where adjustments in terminology would provide a better fit. It would probably be possible to develop a mapping where the word “id” continued to be used in relation to an element of the model that the individual is not aware of, but this would involve different terminology in relation to other parts of the model and would no doubt still generate areas where the mapping was not as simple or as intuitive as one might want. If it involved making even greater changes to the concept of a drive than those Solms proposes, it might well seem that a radical change in the use of the concept of the id was a price worth paying. In any event, these “deep revisions” are not really about mistakes that Freud made – they are stipulations about the particular mapping that Solms wants to use to link the neuroscientific model to the language of psychoanalysis.

Another significant point that comes out in Solms’ more recent work is that he sees neuroscience as studying the mind/brain objectively and psychoanalysis as studying it subjectively, and he believes that there can be useful two-way traffic between the two disciplines. This is an intriguing suggestion, but it is clear that final authority lies with the objective discipline. Psychoanalysis can provide evidence for claims that neuroscience has not yet found ways of testing and in this way it can be a stimulus for neuroscientific research, but the truth is what neuroscience one day manages to establish.

This may seem very reasonable, but its main impact is to mud the waters, since it allows Solms to mix together neuroscientific claims (claims he thinks neuroscience has proved) and psychoanalytic claims which neuroscience has not yet proved and whose neuroscientific meaning has not yet been established. So, we end up with a mixed model which looks more like psychoanalysis than the pure model but only because elements are tacked onto it either for later removal or for later “deep revision”, which may radically change their meaning. The proposed merger of psychoanalysis and neuroscience is a partnership where the ultimate voting rights remain 100% in the hands of neuroscience.