Dysconnection Hypothesis of Schizophrenia

Neuroscientist Karl Friston on split personality, the nature of mental disorders, and the possibility of a therapeutic cure of schizophrenia.

videos | August 10, 2017

Schizophhrenia is a tenacious disease. It’s one of the major psychoses along with psychotic depression. It affects a vast number of people. Understanding of it has grown enormously over the past century, and yet there are some great mysteries about the nature of the disease, its aetiology, what caused it and, indeed, how to cure it. So, typically it’s being regarded as a neurodevelopmental disorder in the sense that it has a characteristic age of onset (usually in the late twenties), more in men than in women, with a very characteristic progression into various flavors of schizophrenia, which people can either diagnose or categorize in terms of particular sources of schizophrenia. Some people prefer to look at clusters of symptoms, and in terms of expression and different dimensions or patterns of symptoms – they themselves fluctuate during the course of the illness.

Dynamic Causal Modelling

Neuroscientist Karl Friston on functional specialization of different brain areas, brain hierarchy, and the connectome

So, originally called «dementia praecox», and much of our understanding of the phenomenology and the nosology comes from the work of people like Bleuler, who, I think, first spoke in an explicit way about the dysconnection implicit in this disorder; spoke of a disintegration of the psyche. By the psyche he meant the internal mental life that somehow fell to pieces, as popularized by the notion of a split brain or a split personality, which is not at all schizophrenia. But that splitting is a hallmark of the mental, psychic aspects that follow from whatever the pathophysiology of schizophrenia entails. There are other notions in the 19th century, speaking again to the notion of a dysconnection, disruption of the organs of connectivity. The white matter tracts that comprise the long axonal processes that mediate the functional integration between different parts of the brain. So Wernicke articulated it as the sejunction hypothesis, where he thought that the pathology of things like schizophrenia was due to a disintegration or a lesion to these organs of connection, the white matter tracks.

So, there’s a long-long history of disintegration and dysconnection of the distributed processing associated with the schizophrenia. Modern-day variants of that hypothesis have emerged under the rubric of a dysconnection hypothesis, borrowing the notion from dysconnection syndromes in neurology. But the key difference between modern versions of the dysconnection hypothesis and classical neurological sejunction-like ideas just to remind that this sejunction notion is an anatomical disruption, a cutting of the wires. Whereas modern-day formulations are a little bit more subtle, these are functional dysconnections.

So if you imagine the brain like a radio with all sorts of specialized components, functionally specialized for doing this bit and that bit (of whatever a radio does in terms of receiving the information and producing outputs), and all the wires as the connections, the anatomical, or sejunction, dysconnection would be a disruption of the wires or failure of the wires to find their proper targets. Whereas the functional dysconnection, the dysconnection hypothesis today, would say «no, it’s your transistors that are broken, it’s the effect of the messages passed down the wires on the local processing that’s broken». So, from the perspective of the brain and systems neuroscience and molecular biology, this would be a subtle but devastating failure of synaptic processing that mediates the functional integration or the connectedness of distributed brain processing. By synaptic failure I mean the mechanisms that allow one brain cell, or neuron, to talk to another. The processes that are in play in gathering messages from pre-synaptic contacts, contacts from either local neurons or distant parts of the brain, and integrating them properly in a context-sensitive fashion, so that each neuron can emit the correct response and pass messages on to the next part of the brain.

Clinical Brain Imaging

Neuroscientist Sylvain Baillet on CT scanning, the way MRI works, and other techniques that revolutionized medicine

So, the picture we have then is that the best explanation, both in terms of the disintegration of the psyche and in terms of a lot of the pathophysiology, the pathology, that you see in brains post-mortem of people with schizophrenia can be tracked down to the level of a failure of synaptic integration. In particular, the sort of failure that endows the architecture with a context-sensitivity, that allows people to recognize what is important, what is not important, to select the right sorts of neuronal information for processing, to build good coherent deep models of their world, so that they can understand it.

So, that contextual aspect speaks exactly to the transistor analogy. It is not the fact that the connection is broken, it’s just not gated properly. In biology that will be known as a neuromodulatory failure. So, immediately we are now in the game of trying to identify, from a functional perspective (under the dysconnection hypothesis), how might you break a synapse, not that you’re removing it, but you’re stopping it contextualizing itself properly, chewing itself to the right strength in this moment given all the neuronal activity that’s going on around that region and other part of the brain. There are lots of mechanisms that are responsible for setting that correct strength, the same way the transistor was set the correct conductance to ensure the right sort of message passing.

We have classical neuromodulatory systems, such as dopamine and serotonin. And dopamine has been particularly interesting, of course, because most efficacious antipsychotic medication that has traditionally been used to treat schizophrenia directly affects dopamine receptors. We also have special receptors – NMDA-receptors – on synapses that have this peculiar nonlinear gating behavior that gives them the capacity to act like a small transistor, to respond in a non-linear way (so that only receptors, the ion channels –they are responsible for controlling – are only in play at certain levels of activational depolarization). So, there are whole theories now that look at the interaction between, say, dopamine receptor activation and NMDA function, and NMDA-receptors are driven by glutamates, so there is a glutamate theory of schizophrenia. In fact, for every neurotransmitter that can affect this neuromodulatory gating there is a synaptic variance schizophrenia, and evidence for, evidence for an abnormality in this particular neurotransmitter system.

There are also other non-neurochemical mechanisms of this gain control, and they speak to ideas about the neural dynamics setting their own excitability. So that if lots of messages arrive at the same time and in synchrony, they will be more potent in eliciting a postsynaptic nervous response than if they arrive asynchronously or slowly. So, that means that there are likely to be disruptions in a person with schizophrenia of this fast synchronized activity – that itself sets the gain control or the balance between excitation and inhibition in the brain that is a key aspect of this gating. So not only do we have the neurophysiological, or neurotransmitter, correlates of attending theories that might explain a dysconnection in schizophrenia, we also have the electrophysiological correlates, that may be more than correlates – they may actually be involving setting the excitability of cortical tissue that mediates the functional integration that is functionally disconnected in schizophrenia.

Finally, from a purely theoretical perspective, what would that mean for a working brain? What would it mean if I wasn’t able to gate, or to select, or to assign weight to two different messages? And the implication of not being able to do that has been most fully developed in the context of the recent hypothesis, like the Bayesian brain hypothesis, like predictive coding. These are just names for a formal understanding of what the brain is doing in terms of processing sensory information. And to cut a long story short, if you break the gain control, if you somehow disconnect, in a functional sense, distributed processing, you will inevitably get false beliefs and false inferences. Your inferences based upon the sensory information will be compromised. False inference, if it’s about sensory percepts, corresponds to hallucinations, inferring something is there that is not there. At higher levels of the brain it corresponds to the delusions, false beliefs. So, to close, we have a graceful way of accommodating the phenomenology, the disintegration, the psyche right from the writings of Bleuler right through to modern-day formulations of the Bayesian brain hypothesis and predictive coding, and speak to false into inferences and disintegration of the psyche that speak to both the synaptic physiology of neuromodulation and the electrophysiology of fast synchronous interactions in the brain.

Behavioral Genetics

Behavioral geneticist Robert Plomin on twin studies, genetic influence of parents on their children, and 1% of DNA that makes people different

So, what does the dysconnection theory mean about future research and possibly therapeutic interventions? If you know that the basic aetiology, the basic pathology lies at the level of synaptic integration, that means that you can focus all your animal models, all your molecular biology models, all your genomics on to synaptic function, and try to identify at what level the synapse is breaking. And, I repeat, it’s likely to be at the interaction between the classical neuromodulator, and, at this point, we believe, things like NMDA-receptor function that have these electrophysiological correlates that are very much bound together. That means that it provides a narrowing of the scope and focus for genomic studies. It also provides an opportunity for a more careful endophenotyping, of different ways that schizophrenia can be expressed, by which I mean, that if you can identify the mechanistic basis of the disorder, you can measure it more accurately and assign different patients to different subgroups; which means that if you are going to understand the correlations between the genetic profile of an individual and the propensity to develop schizophrenia or, indeed, the sort of schizophrenia that they have developed much more accurately and get a much better structure-function understanding that clearly will itself enable a much more rational targeting of different drugs and drug development. One might imagine also, because you have this functional part of the theory, the false inference bit, it also speaks of the possibility of combining drug therapy, which is remedial at the level of this storing or possibly enabling neuromodulatory of gain control mechanisms. At the same time or in conjunction with behavioral or perceptual learning therapies that use experience to reengage the synapses that may have been lost due to a failure of experience-dependent plasticity.