In W v M, a judge concluded that statements made by a woman (M) in a minimally conscious state in the past about not wanting to end up in a residential home or dependent on others should not carry weight in assessing whether artificial hydration and nutrition continued to be in her best interests.1 This judgement has been criticised in the bioethics literature by several commentators for ignoring M's autonomy.2–4 The common argument made by these authors appears to be as follows:

• (P1) Past statements are expressions of an individual's precedent autonomy

• (P2) Precedent autonomy should be respected under the principle of respect for autonomy

• (C) Past statements should be respected

There is a vast literature on advance care planning that argues that the second premise is problematic.5–7 This generally focuses on two issues: (1) the individual's knowledge and level of understanding of the potential health problems they may encounter in the future and (2) the persistence of an individual's personal identity.

In this article, I will take a narrow yet novel approach by examining an inherent assumption contained in the first premise: ASSUMPTION: An individual's past statements represent his/her beliefs, decisions or choices at that time about interventions that may be required or proposed in the event of future health crises.

I will challenge this using a theory of speech acts developed by the philosophers Austin and Searle, a conceptual tool that has previously been used in academic work on consent.8

Speech act theory—things we do with words

In the 1950s, J L Austin argued that when we use language we do not merely describe states of affairs but also perform acts. He coined the assumption that a statement only works to describe the ‘descriptive fallacy’.9 To illustrate his point, he pointed out that there are sentences or groups of words that clearly ‘do’ something when they are uttered. Classic examples would include “I do” at a wedding and “I christen you X” at a baptism. Austin called these ‘performative utterances’.10 If we set aside these more extraordinary examples, we can see that many of the sentences and words used in everyday language also ‘do’ something, and therefore, also have what has been termed ‘illocutionary force’.11 Suppose we say to a partner or friend: “I'm hungry.” Austin's observation is that we are not merely describing a visceral sensation but also often performing a speech act, for example, trying to prompt or persuade our partner to get us something to eat.

Implications for past statements

What are the implications of this descriptive work on the normative question of how past statements should be treated in cases such as W v M? Let us reexamine the type of argument made by several commentators in the literature: ARGUMENT: Past statements are expressions of an individual's precedent autonomy and should therefore be respected under the principle of respect for autonomy

The theory of speech acts shows that past statements do not necessarily communicate an individual's beliefs, decisions or choices. The premise that they are expressions of an individual's precedent autonomy therefore becomes a difficult one to defend. The argument is potentially unsound. Of course, an advocate of this position could counter that past statements should be considered to be expressions of an individual's precedent autonomy, even if this is not always the case. I would suggest this is also problematic as it would result in statements such as “I'd love to be waited on hand and foot” uttered to comfort and reassure a loved one in residential care, being used to decide whether treatment should be withdrawn. I do not know on what grounds this could be justified.

Given the potential consequences, a convincing argument could therefore be made to disregard past statements altogether. However, this may be too strong a position. Even though it may not always be the case, an individual's statements could in fact represent his/her sincere beliefs, decisions or choices about future medical interventions.

The normative question of how to treat past statements is beset by an inescapable epistemic problem. We can never know with any certainty what goes on in an individual's mind when he/she speaks. A speech act's actual sincerity is unknown. However, this is an issue which we deal with every day. Let us consider a consent transaction in clinical medicine. Suppose an individual states, “I consent to Y.” We do not know what they are thinking. However, we infer a permissive mental state from the fact that he/she has said these words to a healthcare professional during a clinical consultation. What type of evidence may be important in assessing past statements?

Tentatively, I would suggest the factors that may be important include an examination of the particular context in which these words are uttered and their consistency with other statements the individual has made and his/her personal life narrative. The type of speech act may also be relevant. Suppose an individual X says to a family member, “Don't put me into a residential home.” It seems reasonable to suggest that this directive speech act may provide stronger evidence of an underlying belief, decision or choice when compared with the assertive “I don't want to be end up in a residential home.”