Commentary on

W.A. Phillips and S.M. Silverstein:
Convergence of biological and psychological perspectives on cognitive coordination in schizophrenia
Behavioural and Brain Sciences 26 (2003), 65-137.

A wide-spectrum coordination model of schizophrenia

Hendrik Pieter Barendregt
Nijmeegs Instituut voor Informatica en Informatiekunde
Box 9010
6500 GL Nijmegen, The Netherlands

telephone: +31.24.365 2727


The paper under review presents a model for schizophrenia extending four levels of abstraction: from molecules, cells, cognition to syndrome. An important notion in the model is that of coordination, applicable both to the level of cells and cognition. The molecular level provides an `implementation' of the coordination at the cellular level, which in turn underlies the coordination at the cognitive level, giving rise to the clinical symptoms.

The presented model of schizophrenia can be depicted as follows.

\begin{displaymath}\fbox{NMDA $\downarrow{\;\Rightarrow\;}$neur. coord.$\downarr...

This requires some explanation from the following dictionary.

NMDA&N-methyl-D-aspertate glutamate...
...ow$& $X$ decreases\\
$X\uparrow$& $X$ increases

In somewhat more detailed terms the model states the following. If the activity of NMDA glutamate receptors in the cortex is below normal, then neural coordination within and between cortical regions is decreased; this in turn implies decreased cognitive coordination, such as disambiguation and dynamic grouping; this then will be the direct cause of the symptoms of schizophrenia, like impairments of perception, pre-attentive sensory gating, selective attention, working memory and long-term memory. The authors choose to focus on disorganization.

The way the authors come to their model is as follows. NMDA antagonists cause schizophrenia-like symptoms; schizophrenia implies impaired cognitive coordination and vice versa; neural coordination is behind cognitive coordination. The model postulates that the NMDA glutamate channels provide a control for the neural coordination. The rationale behind this is that the NMDA receptors are voltage-gated, i.e. depend on both the ligand and the right voltage to be opened. So they may be used indeed to coordinate processes (the voltage-gating me be used for calibrations of the receptor).

One virtue of the model is that it is wide-spectrum. It ranges from a molecular mechanism via cellular phenomena, via cognition to psychiatric symptoms. The model makes predictions about patients suffering form schizophrenia: there is impairment of global, but not local motion perception; high frequency rhythms (gamma) will be reduced. This implies that the model is falsifiable.

The main virtue of the model is its emphasis on coordination, interaction. In computer science a notion and theory has emerged, not yet known to the authors, that seems relevant here: communicating systems, see [4]. Although everything happening in a computer may be described by fluctuating bits, the theory of communication forms a convenient level of abstraction. Some bits encode meaningful information to be used later, other bits represent actions that are relevant at the very moment. An interacting communication, the most fundamental concept in the mentioned theory, needs two half-acts, waiting for the other half to be present simultaneously (like two persons that want to shake hands)1. This all may be useful for a thorough theoretical underpinning of the way in which coordination is implemented by NMDA glutamate channels.

Although a single model for schizophrenia is presented, this does not imply that it is a homogeneous condition. The reason is that there are many ways in which coordination can be impaired. Also the effects can vary in severity. The authors give several examples. Moreover this also is clear from the computer science theory of communicating systems, mentioned above.

The authors mention similarity of their model with many other theories, but not with all theories and not in all aspects. The theories putting forward as cause the disconnection between cortical regions, is mainly similar, apart from the fact that it does not speak about coordination within regions and focus on long term, i.e. learning, effects, see [1,3]. The authors do focus in their model on the cortex, but mention that other brain regions will be involved. A paper not mentioned by the authors in which this is described is [2]. In this reference a model of the pathogenesis of schizophrenia is given in which the drive and guidance mechanisms in the brain (specific brain regions are mentioned) are said to be underdeveloped (in the terminology of the paper under review: do not coordinate well). Such intentional aspects are fitting well with the model of the paper under review.

The paper ends by stating many open questions. One puzzle that has been ignored is the claim in [5] that some of the schizophrenic patients get ``weller than well''.

Footnote. 1A typical example of a communicating process is a vending machine. It has a slot for coins and one button for coffee and one for tea. The process of the machine is

\begin{displaymath}\mbox{M=want\_coin.(ready\_tea-button + ready\_coffee-button).M.}\end{displaymath}

This means that the machine is waiting for a coin and after that for either a push on the tea or the coffee button. Here ``.'' stands for sequential composition and + for choice. That in the definition of M there is again an M on the right hand side is due to the fact that we like the machine to keep operating. A human that regularly wants to use the machine has as process

\begin{displaymath}\mbox{H=put\_coin.(push\_tea-button + push\_coffee-button).H}.\end{displaymath}

Now the interaction of the human with the machine is denoted by $\mbox{H$\Vert$M}$. Provided that we postulate that there are communications c such that


and that non-communicating processes (like c(push_coffee-button,ready_tea-button) are abstracted away, we obtain

accept\_coin.(pours\_tea + pours\_coffee).(H$\Vert$M)}\end{displaymath}

and that is indeed the outlook on the world from the point of view of such a vending machine (we left out considerations that they need a refill and that water and energy are available in unlimited quantities). We see the difference with ordinary algorithmic programming. That is directed towards termination. Programming a process is often directed in an interactive environment to unlimited continuation.

The theory of communicating systems carefully describes processes with a global control, versus ones with a local distributed control without global knowledge.

The theory of mobile systems (also see [4]) goes beyond that of communicating ones. The intended model in ITC is that of mobile telephones or web-pages with links. Here the number of action channels is variable and a communication may create a new channel between other processes. This theory may model very well the way in which cells communicate. In some cases there is in a cell no receptor for a certain transmitter T, but there is for another transmitter T'. Reception of T' will cause that the DNA-code for the receptor for T is read from the genome and brought to expression so that T can be received.


[1] Dolan et al. (1999) Abnormal neural integration related to cognition in schizophrenia, Acta Psychiatrica Scandinavica 99, 58-67.

[2] van Hoof, J.J.M. (2002) The abnormal development of drive and guidance mechanisms in the brain: the pathogenesis of schizophrenia, Acta Neuropsychiatrica 14, 134-146.

[3] Friston, K.J. (1999) Schizophrenia and the disconnection hypothesis. Acta Psychiatrica Scandinavica 99, 68-79.

[4] Milner, R. (1999) Communicating and mobile systems: the $\mbox{${\pi}$ -Calculus}$, Cambridge University Press.

[5] Menninger, K., M. Mayman and P. Pruyser (1963) The vital balance, The Viking Press.

Henk Barendregt