Tuesday, 9 August 2016

Delusions and Language: the Forgotten Factor

This post is by Wolfram Hinzen, who is research professor at ICREA (Catalan Institute for Advanced Studies and Research), currently investigating language as an aspect of human nature, cognition, and biology at the Grammar and Cognition Lab. In this post, he summarises his recent paper "Can delusions be linguistically explained?", co-authored with Joana Rossello and Peter McKenna and published in Cognitive Neuropsychiatry.

Wolfram Hinzen

Explanations of delusions often revolve around meaning and knowledge. Indeed the philosopher and psychiatrist Karl Jaspers (1959) famously argued that all delusions could be understood as a radical transformation in the awareness of meaning, which became immediate and intrusive attribution. How such an abnormality might give rise to statements such as ‘I am Jesus’, ‘the Royal family are stealing my inventions’ and in extreme cases ‘I have fathered 7,000 children’ or ‘I have invented a machine to run the whole solar system powered by milk’, was never very clear in this account. Nor for that matter do contemporary psychological theories on semantic memory, which contains knowledge about the meaning of words (ie the lexicon) as well as all other knowledge about the world, provide any obvious way via which dysfunction could give rise to the kind of statements deluded patients make.

There is however another level of meaning, whose existence has long been recognized, particularly in philosophy. This is propositional meaning, the kind of meaning that arises from structuring of lexical-semantic information when the level of complexity of full sentences is reached. Some neuro-cognitive system needs to generate propositional meaning, and according to the un-Cartesian hypothesis (Hinzen, Sheehan, 2013), this system is the system that retrieves lexicalized concepts stored in long-term semantic memory and puts them into grammatical configurations. As a result, a new kind of meaning arises that lexical concepts as such do not carry: propositional and referential meaning. Grammar, according to this proposal, is what ultimately enables the ‘world as known’, the world as it appears in the format of propositional knowledge.

Furthermore, it is only at the level of propositional or grammatical meaning that considerations of truth and falsehood arise. Single words or phrases, like ‘man’, ‘the man’ or ‘killed Bill’ cannot be true or false. However, ‘The man killed Bill’, as uttered in context, will be. The quality of falsity that delusions have is clearly a function of their propositional (not their lexical) meaning.

Peter McKenna
In the case of ‘I am Jesus’, this sentence as uttered in mentally healthy individuals can only carry non-literal (e.g. metaphorical) meaning, or perhaps it is the utterance of an actor pointing to a character he plays on screen (i.e. him in the Jesus-role). But the patient who says this does not think he plays a role. Listening to his utterance it is hard to tell whether he is referring to Jesus or to himself – a confusion which I and my colleagues Joana Rosselló and Peter McKenna have recently argued is essentially linguistic in nature, reflecting a collapse of reference and predication.

In bizarre or fantastic delusions such as ‘I have a wine glass in my stomach’, the breakdown in propositional meaning becomes more flagrant – such statements only have the appearance of propositional meaning, but do not carry any information about the world. The patient is not referring to a possible if unlikely scenario such as a freak accident; instead, stomach and wine glass are simply juxtaposed and presented as a viable proposition.

What about delusions such as ‘the mafia are trying to kill me’, which is a possible if in normal circumstances very unlikely scenario? Here we argue that propositional meaning always takes place in a frame where objects and events are located in relation to speakers, hearers, and a context of other statements that have taken place before: any statement always follows upon other statements, which the persons conversing have both witnessed and mutually know they have witnessed. This ‘frame’, however, is constructed out the grammatical meaning of previous utterances; in fact, it depends completely on this. However, in the case of a patient who believes he is being persecuted the proposition is uttered in the absence of such a frame; the proposition does not refer to events in the world, as referred to in previous assertions and instead it refers only to mental objects, to the thinking process of the patient itself.

Joana Rossello
We argue that another propositional level linguistic abnormality is also characteristic of delusions, failure of embedding. Delusions by their very nature cannot be negated, questioned, or subordinated, e.g. ‘I sometimes feel like I am Jesus’. If a deluded patient were able to subordinate (embed) a delusional utterance in this fashion, it would stop qualifying as a delusion clinically. Things that cannot be embedded cannot be questioned, thus becoming hardened truths that cannot be justified or criticized, reflecting no learning and allowing none.

Being able to place delusions in a linguistic framework may be heuristically useful – it puts them on the same conceptual footing as the two other classes of positive or productive psychotic symptoms, formal thought disorder (the incoherence of speech seen in some patients with schizophrenia) and auditory hallucinations. The former symptom is linguistic in the sense it is a disorder of communication, and various kinds of language abnormality have been shown to contribute to it (McKenna, Oh, 2005). The vast majority of auditory hallucinations take the form of hearing voices and so are in an important sense also linguistic.


  1. My response to the post. Delusions are not linguistic, they're discursive.


  2. I would argue that the distinction between "bizarre" and non-bizarre delusions is a false one and collapses under scrutiny. In the examples you give above ("I am Jesus" and "I have a wine glass in my stomach") the statements have a meaning that we readily understand. In the first case, assuming we don't think that Jesus is alive or present, the problem is one of reference. If someone were to say he were being attacked by a satellite that turned countries into deserts (a problem of reference), if I were able to get him to accept that instead it was a (real, verifiable) Chinese satellite, his delusion would not have meaningfully changed, just recruited a new element. Interestingly, delusional patients do often "trail off" and start making incomprehensible (a-grammatical, nonsensical) utterances when asked to elaborate about the delusion, and the structure of such utterances (if any) would be of interest. Also, I plan to try work with delusional patients and see if they are able to grammatically subordinate delusional claims.

  3. Thanks for your comment - I find interesting what you say about 'trailing off'. We are currently investigating such 'agrammatical' utterances in the context of thought disorder, and it would be very interesting to see them in the context of elaborating a delusion. It is hard, though, to find objective linguistic criteria for annotating such agrammatical speech. On the capacity to grammatically subordinate delusional claims, again I would be keen to hear your observations. I suppose that the more such subordination is seen, the less 'fixed' the delusion is and the more it can be dealt with. On your first point, I would reply as follows: Reference is not a word-level phenomenon but a grammatical one. Ultimately noun phrases have reference only when occurring in sentences, or even in discourse (which relates to the first post above). So in the Jesus delusion, the problem is not merely one of the reference of the word Jesus in isolation, but of deictic embedding of all NPs together in their grammatical context, and specifically the coordination of the grammatical first and third persons in this utterance (which is what creates an impossibility). In your satellite example, the situation is similar, though there is no overt first person. If indeed the satellite was real, then I would say that the delusion has meaningfully changed - namely in terms of deictic embedding of the NP in its sentential and cross-sentential context. Language would have found its normal deictic anchors. One could define a notion of meaning that is independent of reference, but in natural language, I would say that things don't work that way, and meaning is inherently integrated with reference.

  4. https://www.academia.edu/7949715/The_Generation_of_Psychosis_A_Pragmatic_Approach


Comments are moderated.