Establishing the accuracy of self-diagnosis in psychiatry

1. Introduction

“Autistics know autism better than anybody else” (Enright, Citation2021). This quote comes from a blog post by a self-diagnosed autistic individual. They use the notion that autistic people know autism best to justify self-diagnosis. In this article, I consider what arguments are needed to understand the claim that “autistic individuals know autism best” as entailing that self-diagnosis has high accuracy.

Self-diagnosis in psychiatry is where individuals diagnose themselves rather than rely upon official diagnosticians to supply a psychiatric diagnosis. The legitimacy of self-diagnosis has been supported and challenged by individuals who have been diagnosed by official diagnosticians. For example, in relation to autism, the “subject of self-diagnosis is a perennial one here [an internet forum for autistic individuals] and generally causes heated discussion” (unnamed autistic individual quoted in Sarrett, Citation2016, p. 27) and “[w]hether one can self-identify as autistic is a fierce topic of debate… within the autistic community” (Hens et al., Citation2018, p. 3). One debated issue is the degree to which someone can accurately establish whether they are or are not an instance of a particular diagnosis, such as whether they are or are not autistic. In this article, I investigate what needs to be established for self-diagnosis to be accurate.

I focus upon establishing what arguments are needed to take self-diagnosis as accurate. Very little philosophical work has been done on the accuracy of self-diagnosis, so my aim is to establish which argument advocates of self-diagnosis need justify and which argument opponents of self-diagnosis need refute. Additionally, I outline four different approaches to justify the accuracy of self-diagnosis and show that each approach justifies the accuracy of self-diagnosis on different grounds. This will help prevent advocates of self-diagnosis formulating an argument which is required for one approach and using it to justify a different approach to self-diagnosis. Similarly, it will help prevent opponents of self-diagnosis from countering arguments used to justify one approach to self-diagnosis and then taking this as impacting other approaches to self-diagnosis. I will outline how there are multiple steps to justifying each of the different approaches. Some of those steps are applicable to more than one approach, but each approach also requires specific arguments that are not applicable to other approaches. Therefore, good arguments for one approach would not then justify one of the other approaches.

There are three areas I will not analyze in this article. Firstly, having established what arguments are required to see self-diagnosis as accurate, I will not then assess the prospects for supplying those arguments. I identify four different approaches to seeing self-diagnosis as accurate, and a detailed analysis is needed to show that different arguments are required to justify each approach. Secondly, I will not consider whether self-diagnosis is legitimate even if self-diagnosis is inaccurate. There might be good reason for thinking self-diagnosis is justified even if a significant minority or a majority of self-diagnosing individuals inaccurately self-diagnose, but I shall not investigate this possibility. Thirdly, even if self-diagnosis is accurate, separate ethical arguments are needed to claim that self-diagnosis is justified. I will assume individuals have an ethical right to self-diagnose if they cannot access an official diagnosis or if official diagnosticians will not accurately diagnose them. This seems to mirror most arguments about self-diagnosis whereby the ethical principle is rarely challenged and the debate focuses upon accuracy.

2. Terminology

I have seen a variety of terminology used when self-diagnosis is being discussed. Terminology includes the reliability, the validity and the accuracy of self-diagnosis. I avoid the terms reliability and validity because they have technical meanings in psychiatry which are not fully applicable to self-diagnosis.

Reliability relates to the agreement between multiple clinicians about which diagnostic criteria are being met by a particular individual (Aragona, Citation2014; Spitzer, Citation1978). In contrast, self-diagnosis typically relates only to one person, the self-diagnosing individual. As such, reliability in the technical psychiatric sense does not seem applicable to self-diagnosis because there is no issue of agreement between multiple individuals. We could extend notions of reliability to include the self-diagnosing individuals and ask what level of agreement there is between the self-diagnosing individuals and official diagnosticians. This, however, would be hypothetical since most self-diagnosing individuals never get an official assessment. We could run a hypothetical thought experiment by considering the likely agreement between the self-diagnosing individuals and official diagnosticians. However, people often self-diagnose because they expect official diagnosticians to not diagnose them. This suggests that the level of agreement will be low and so, in this technical meaning, self-diagnosis would have low reliability.

Validity also has a technical usage in psychiatry. At its most basic, validity relates to correlations whereby people with a psychiatric diagnosis can be correlated with other factors such as long-term outcome and genetics (Robins & Guze, Citation1970; Solomon, Citation2022). Additionally, some take sufficiently high correlations as indicating the reality of the diagnosis or that the diagnosis is a biomedical entity (Kendell & Jablensky, Citation2003; Pies, Citation2008). Whether this is or is not a good way to understand psychiatric diagnoses depends upon what we take psychiatric diagnoses to be (an issue I have discussed elsewhere (Fellowes, Citation2022)). However, I will later outline ways in which self-diagnosing individuals can reject current DSM criteria and instead self-diagnose on alternative criteria without aiming to justify this on appeal to correlations, reality, or biomedical entities.

As such, the technical meaning of reliability and validity in psychiatry does not map well onto questions of whether self-diagnosis is accurate. Instead, I will talk of the accuracy of self-diagnosis. I understand accuracy in the following two senses. Firstly, if the DSM criteria for autism are accepted, then can the self-diagnosing individual accurately establish if they do or do not meet the diagnostic criteria? Secondly, if the DSM criteria are taken as not accurately describing autism, then can a self-diagnosing individual produce a more accurate description of autism when self-diagnosing?

3. Arguments for and against the legitimacy of self-diagnosis

The basic idea of self-diagnosis is that an individual relies upon their own judgment about whether they have a particular diagnosis rather than relying upon the views of designated medical professionals. The person who receives the diagnosis is also the person who makes the diagnosis. This contrasts with official diagnosis whereby a medical community designates official diagnosticians to provide official diagnoses. An official diagnosis is placed on a medical record and is typically required to access support services and receive state benefits.

The phrase self-diagnosis is potentially misleading. Firstly, the self-diagnosing individual may consult other individuals during the process of self-diagnosing, such as friends, family, online communities, or even medical professionals. The key point is that a self-diagnosing individual does not hold an assessment with official diagnosticians or the self-diagnosing individual rejects the views of official diagnosticians if they are assessed. Secondly, self-diagnosing should be seen as distinct from suspicion of diagnosis. Many people who are waiting for an official diagnosis will tentatively or strongly suspect they meet the diagnostic criteria prior to an official assessment. For my purposes when I refer to self-diagnosis I mean an individual who is not currently waiting for an official diagnosis.

I shall explore self-diagnosis by considering self-diagnosis in autism. Self-diagnosis seems quite a restricted phenomenon, primarily occurring in relation to a small number of diagnoses. For example, Sarrett notes that self-diagnosis seems to be “relatively unique to autism within… psychiatric, developmental, and intellectual disabilities” (Sarrett, Citation2016, p. 26). For empirical evidence of this claim, Sarrett consulted one internet forum which is intended for autistic individuals and another internet forum which is intended for schizophrenic individuals. She found regular discussions of self-diagnosis on the forum for autistic individuals, whereas on the forum for schizophrenic individuals she found “self-diagnosis not to be an issue” (2016, p.26). Additionally, to my knowledge more academic studies have been done on autism compared to other diagnoses. As such, there is a stronger evidence basis when discussing autism compared to any other diagnosis (see however, Chan & Sireling, Citation2010; Giles & Newbold, Citation2011 in relation to bipolar disorder). It is certainly possible that self-diagnosis is more common in other diagnoses than is represented by the empirical literature. I see no reason why my argument will not be applicable beyond autism, but I cannot rule out the possibility that focusing upon autism will mean some parts of my argument cannot be generalized to other diagnoses.

Despite having a stronger evidence basis than other diagnoses, there are still relatively few studies on self-diagnosis of autism. McDonald writes that, “[l]ittle is currently known about this self-diagnosed group of [autistic] individuals” (McDonald, Citation2020, p. 15). To my knowledge, there are two empirical studies which focus entirely (Lewis, Citation2016; Sarrett, Citation2016) or significantly (Lewis, Citation2017; McDonald, Citation2020) upon self-diagnosis of autism. As such, I have limited empirical data on the motives for self-diagnosing.

There are three main strands to justifying self-diagnosis. Firstly, an official diagnosis is inaccessible. Secondly, being diagnosed can be harmful or not helpful. Thirdly, that the official diagnostic process is inaccurate whereas self-diagnosis is accurate.

In relation to inaccessibility, there can be long waiting times to see official diagnosticians (Lewis, Citation2016, Citation2017, p. 578; McDonald, Citation2020, p. 15; Sarrett, Citation2016, p. 31). Seeing official diagnosticians can sometimes have a substantial financial cost (Lewis, Citation2016, Citation2017, p. 578; McDonald, Citation2020, p. 15; Sarrett, Citation2016, p. 31). Finally, some individuals struggle to navigate the healthcare system (Lewis, Citation2016).

In relation to diagnosis being harmful or not helpful, some individuals (often based upon past experience) believe that medical professionals may harm them (Lewis, Citation2016, Citation2017, p. 578). There is concern that being diagnosed as autistic could harm future employment prospects (Lewis, Citation2016, Citation2017, p. 578; Sarrett, Citation2016, p. 31). Finally, the lack of available support services raises questions about whether it is worth actually getting officially diagnosed (Lewis, Citation2016, Citation2017, p. 578).

In relation to the diagnostic process being inaccurate, there is concern that official diagnosticians will mistakenly claim the individual does not meet the diagnostic criteria of autism (Lewis, Citation2016, Citation2017, p. 578). The danger is that official diagnosticians hold particular stereotypes of autism and fail to realize that autism can manifest in ways which exceed the boundaries of those stereotypes. As an example, intense interest in focused topics is typically considered a symptom of autism. Popular stereotypes portray autistic people as deeply focused upon transport or computers. As such, official diagnosticians might be looking out for such interests when assessing if someone is autistic. However, these stereotypes seem more applicable to how some, though not all, western autistic boys manifest intense interests. In contrast, autistic girls are more likely to have an intense interest in ponies or boy bands (Moseley et al., Citation2018, p. 2). Official diagnosticians who take intense interest to only manifest in a limited number of stereotypical ways have a significant chance of missing detecting intense interest in individuals who manifest them in ways outside of those stereotypes. There is a danger that official diagnosticians will fail to recognize that an individual is autistic. Also, some individuals find describing their symptoms to official diagnosticians difficult (Lewis, Citation2016). As such, whilst the individual may understand which symptoms they exhibit and how they exhibit them, they may be unable to communicate this to official diagnosticians. Additionally, an autistic person might have little opportunity to present the manner in which they are autistic in a meeting with official diagnosticians (Sarrett, Citation2016, p. 31). They manifest autism in complicated and nuanced manners which vary depending upon the particular real-world situation they are in. In contrast, they will be far removed from those real-world situations when assessed by official diagnosticians, so will have less opportunity to portray how they actually manifest autism. Finally, autistic individuals sometimes mask (Livingston & Happé, Citation2017, p. 732). This is where they effectively try and pass as normal. This typically leads to a suppression of behavior that is present in diagnostic criteria.

Arguments that official diagnosis is inaccurate are typically supplemented with arguments that self-diagnosis is accurate. Autistic individuals have lived experience of being autistic whereas a non-autistic individual only has access to the experience of observing autistic individuals. As such, an autistic person is taken as having more direct access to autism than official diagnosticians have. Therefore, an autistic person is better able to assess that they are autistic compared to official diagnosticians (Sarrett, Citation2016, p. 30). This intuition is expressed in the quote which I used to start this article that is from a blog by an individual who is self-diagnosed with autism: “Autistics know autism better than anybody else” (Enright, Citation2021). This means that self-diagnosis can be as accurate or more accurate than an official diagnosis. Also, in relation to individuals being only able to manifest their characteristics in a limited number of ways when meeting official diagnosticians, the autistic individual is aware of those manifestations taking place in contexts outside the assessment, so they have more knowledge of how they are autistic compared to official diagnosticians who cannot see those manifestations taking place in everyday settings.

Some counter arguments for self-diagnosis have been identified in the literature (Sarrett describes rather than endorses these counterarguments). The main concern is that a self-diagnosing individual lacks the training that official diagnosticians have (Sarrett, Citation2016, p. 27). The concern is that a level of knowledge and training is required to accurately diagnose autism. There is also a concern over potential bias when self-diagnosing. An individual who self-diagnoses might do so by falsely interpreting some of their characteristics as being instances of autism. They are looking for characteristics of autism and falsely end up seeing what they expect to see (Sarrett, Citation2016, p. 28). Finally, there is a belief that some people who self-diagnose as autistic are insufficiently disabled to be autistic. There is the concern that any individual who was sufficiently disabled to be autistic would have already been spotted by medical professionals. Therefore, some autistic individuals believe that self-diagnosing individuals cannot be autistic (Sarrett, Citation2016, p. 28).

There are two fundamentally different approaches to the accuracy of self-diagnosis which need to be distinguished between. In one approach, the individual intends to self-diagnose based upon official diagnostic criteria. They do not challenge official diagnostic criteria but consider themselves to be an accurate judge of whether they meet the official diagnostic criteria. Additionally, they might consider themselves a better judge than official diagnosticians. For example, as discussed above, official diagnosticians might associate symptoms on the diagnostic criteria with a limited range of stereotypical manifestations. In contrast, a self-diagnosing individual can draw upon their lived experience to realize that their symptoms manifest in particular ways which official diagnosticians fail to realize. In a second approach, the self-diagnosing individual rejects current diagnostic criteria. They believe that someone can be autistic despite not meeting the diagnostic criteria in the main official diagnostic manuals, the DSM (Diagnostic and Statistical Manual of Mental Disorder) and the ICD (International Classification of Disease). They believe current diagnostic criteria are flawed and are not an accurate guide of establishing who is autistic. As such, official diagnosticians being able to accurately establish that an individual does not meet the official diagnostic criteria for autism does not then entail that official diagnosticians are an accurate judge of who is autistic. Accurate diagnosis of autism requires modification of the diagnostic criteria of autism present in the DSM and ICD. Some arguments required to justify self-diagnosis when accepting official diagnostic criteria have no relevancy when rejecting the diagnostic criteria and visa versa. I will distinguish between these situations in the following discussion.

4. The accuracy of official diagnosis

As mentioned above, one significant motive for self-diagnosis is the belief that official diagnosis can be inaccurate. I do not intend to take a stand on this matter in this article for the following reasons. Firstly, debates about the reliability (in the technical sense described above) of psychiatric diagnoses have been going on for many decades (Kendell, Citation1975, see Aragona, Citation2015 and; Solomon, Citation2022 for history) and there is disagreement about how high the reliability of the DSM is (Chmielewski et al., Citation2015; Vanheule, Citation2017). I have little to add to this debate. Many readers will come to this article with preexisting views on the accuracy of official diagnosis. I likely have no greater ability to change anyone’s mind on this issue than any of the literally hundreds of existing articles and books on the topic which have already been published. This is especially true given that I have no committed views upon this topic. Secondly, even if official diagnosis can be inaccurate that does not then prove the accuracy of self-diagnosis. It could be the case that official diagnosis can be inaccurate and self-diagnosis is even more inaccurate or that both are equally inaccurate. We need to establish the level of accuracy of self-diagnosis to make a comparison with official diagnosis regardless of whether official diagnosis is considered to have high or low accuracy.

However, given that advocates of self-diagnosis are often concerned over official diagnosticians holding inaccurate stereotypes of autism, it is briefly worth commenting upon this specific issue. There is good reason to believe that at least some official diagnosticians will be aware that some stereotypes of autism are misleading. For example, in relation to autism, common concerns that advocates of self-diagnosis hold are actually addressed in the DSM 5. Advocates are concerned that official diagnosticians lack awareness about autistic people masking and about under-diagnosis of autism in ethnic minorities and women. However, the DSM 5 entry for autism explicitly mentions that autistic people can mask (APA, Citation2013, p. 56), that autism can be under-diagnosed in ethnic minorities (APA, Citation2013, p. 57), and that autistic women may present differently by exhibiting “subtle” (APA, Citation2013, p. 57) social skills difficulties. The degree to which official diagnosticians take note of these points, spend time considering them, and ideally do further reading about them is something that, to my knowledge, we have no data on. However, given that the DSM 5 itself mentions these issues, it seems credible that at least some official diagnosticians have at least some awareness of these issues.

Also, official diagnosis can involve multi-disciplinary teams. For example, according to the National Institute for Health and Care Excellence (NICE) in England and Wales, “it is recommended that the assessment should be done by a multidisciplinary team” (Scattoni et al., Citation2021, p. 4130). Different individuals on the team will be specialists in different areas of autism, and they will interact with the person being assessed in different ways. As such, it seems plausible that diverse medical backgrounds and different styles of observations would help with recognizing that autism can occur in ways beyond a limited set of stereotypes. An advocate of self-diagnosis could argue that all or most of the members of the multi-disciplinary team hold misleading stereotypes, but this claim would require showing that the stereotypes of autism are not just prevalent in one or a few disciplines but actually prevalent in many different disciplines. I am not aware of advocates of self-diagnosis having shown this and I think advocates of self-diagnosis should prioritize providing evidence for this. However, even if multi-disciplinary teams do increase the accuracy of autism diagnoses, it should be noted that official diagnosis often does not involve multi-disciplinary teams. In relation to autistic people who were officially diagnosed, a study of 11 European countries found less than 45 percent saw a multi-disciplinary team (Scattoni et al., Citation2021, p. 4136) whilst a study in Australia found that 66 percent only saw a single clinician (Huang et al., Citation2022, p. 2991).

Additionally, official diagnosis can involve diagnostic instruments like DISCO, ADOS, and ADI-R (Evers et al., Citation2021). Ideally, diagnostic instruments would reduce the level of subjectivity involved in official diagnosis. When using diagnostic instruments, judgments about whether the diagnostic criteria made by the official diagnostician or the multi-disciplinary team are being supplemented by standardized diagnostic instruments which have been proven to be highly reliable (in the technical sense discussed above). In this regard, the diagnostic process can be seen as involving less of a subjective judgment than advocates of self-diagnosis believe. However, the key problem here is that those diagnostic instruments might have only been shown to be accurate when we consider a relatively narrow range of presentations, and they are not accurate at detecting a wider range of manifestations of autism. Whilst I will not assess this claim, it is worth mentioning that this claim has been made by official diagnosticians (Hayes et al., Citation2022, p. 494) and academic researchers (Harrison et al., Citation2017, p. 2771) in articles that do not discuss self-diagnosis, meaning it is not just advocates of self-diagnosis who are concerned about this.

5. Identities and accuracy

The whole question of accuracy might be irrelevant if psychiatric diagnoses are seen as identities. I now address this possibility. Psychiatric diagnoses are traditionally understood to fall under the domain of medicine. A psychiatric diagnosis signifies the presence of a disorder, disease, or illness. This then entails legitimacy for treatment, care, medication, or therapy. However, movements like mad pride and neurodiversity partially or fully reject this. Rather, a psychiatric diagnosis like autism is primarily or entirely a difference and not a disorder. Additionally, psychiatric diagnoses can also be seen as an important aspect of personhood. To remove autism would be in some sense to remove the person. There is not a person plus autism, rather, autism is an integral aspect of who someone is. This raises the possibility of thinking of psychiatric diagnoses more as identities than diagnoses. Just as someone does not self-diagnose themselves as homosexual or trans but rather identifies as these, so too we might say people should identify as autistic rather than self-diagnose as autistic. However, I do not believe that thinking in terms of identity rather than diagnosis makes a difference to my argument.

At least in relation to autism, many advocates of self-diagnosis use the phrase self-diagnosis whilst simultaneously thinking that autism is not a disorder. As such, they seem to think that diagnosis is separable from disorder that people can be legitimately diagnosed despite not being disordered. Additionally, they typically think autism can or even should form part of a diagnosed person’s identity despite seeing autism as a diagnosis. As such, many activists do not see a tension between diagnosis and identity. Whilst I also do not think this is problematic, I now explore the consequences for accuracy if we think something being an identity means it cannot be a diagnosis.

It could be argued that if psychiatric diagnoses are just identities, then the issue of accuracy does not arise. To be autistic is to identify as autistic. If autism is nothing more than an identity, then the only criterion for being autistic is identification. As such, self-diagnosing as autistic means you automatically meet the sole criterion for being autistic. We could say that questions of the accuracy of self-diagnosis simply do not arise, or we could say that the accuracy is always one hundred percent. However, I now raise multiple problems with this approach.

Firstly, this approach seems to deny the possibility that someone can be mistaken about their identity but plausibly people can be mistaken about some types of identities. There is the possibility of delayed identification. Someone might identify as homosexual long after they regularly had same sex attraction, especially if they resist the notion that they are homosexual. Similarly, it seems plausible to say that someone is autistic either from birth or within a few years of birth even if they only get diagnosed or self-diagnose when aged 40. In both cases, the individual seems to be an instance of the identity long before they adopt the identity. Also, there is the possibility of being purposefully misled. An individual might grow up identifying as Scottish based upon being told they were born in Scotland but later found out they had been adopted and were actually born in England. They consequently think they should have always identified as English and were incorrect to identify as Scottish. Similarly, imagine that a child is diagnosed as autistic, but their parents tell them they have been diagnosed with a different diagnosis that the parents consider less stigmatizing. The child identifies as that other diagnosis but finds out much later that they were diagnosed as autistic. That person might think they mistakenly identified with that other diagnosis and actually should have identified as autistic. Finally, there is the possibility of accidental misunderstanding. Imagine that someone identified as a Trotskyist on the basis of holding strong left-wing principles, whilst also rejecting Stalinism. By rejecting Stalinism, they default to a “Stalin is bad, therefore Trotsky is good” principle. Much later, that individual reads about Trotsky and comes to the opinion that Trotsky as leader of the Soviet Union would have likely also lead to the deaths of millions. As such, they consider themselves mistaken to have identified as a Trotskyist. Similarly, imagine someone self-identified with one diagnosis because it seemed to fit better than the other diagnoses they were aware of, then they later learn of another diagnosis which they feel fits themselves. Consequently, they self-identify with the diagnosis they have recently learnt about and consider themselves to be mistaken to have self-identified with the earlier diagnosis. There seems to be a sense in which we can be wrong about some identities which then raises questions about the accuracy of self-identifying with psychiatric diagnoses even if psychiatric diagnoses are actually understood as purely identities.

Secondly, Robert Chapman has also produced strong arguments against seeing autism as simply an identity. He mentions that “autism includes a small subset of individuals who have not yet been able to learn the level of language necessary to identify with other autistic individuals at all” (2020, p.809). Looking at autism purely as an identity consequently “would exclude those who clearly are autistic but are not currently able to identify as such” (2020, p.809). That being, if someone who is currently considered autistic but lacks, for whatever reason, the capacity to identify themselves as being autistic then they are not autistic if autism was purely an identity. This has the obvious consequence of potentially denying support and community to people who have relatively high support needs.

Thirdly, one motive for seeing autism as an identity is denying that autism is a scientifically legitimate entity (Chapman (Citation2020) outlines but does not endorse this argument). Autism covers a heterogeneous collection of symptoms and causes, there are no clear boundaries between autism and other diagnoses and no clear boundary between autism and the general population. This can be done to show that autism is not something that we discover, rather, it could be argued we have to decide what autism is. As such, to be autistic is to simply decide that you are autistic. I outline in sections 9.3 and 9.4 that we can see psychiatric diagnoses as the product of a decision-making process, but this is compatible with there being better and worse ways of deciding which psychiatric diagnoses there are. As such, someone could identify as being autistic through having made a poor decision about where the boundaries of autism lie. I later suggested that this would mean that they were mistaken to think of themselves as autistic.

6. Clarifying the argument

I start by making the basic justification for self-diagnosis more precise. I start by considering the claim that self-diagnosis is accurate because self-diagnosing individuals have self-knowledge of the diagnosis. This discussion is applicable to both accepting and rejecting the DSM and ICD diagnostic criteria when self-diagnosing.

7. Self-knowledge

All approaches to justifying the accuracy of self-diagnosis require making arguments relating to self-knowledge. I will outline the required arguments here. Note that justifying these arguments does not then by itself show that self-diagnosis is accurate. Rather, these arguments relating to self-knowledge need justifying and then other additional arguments also need justifying.

8. Self-diagnosing when following the DSM and ICD

In this section, I address self-diagnosing individuals who accept the diagnostic criteria present in the DSM and ICD. If they self-diagnose themselves as autistic, then they take themselves as being an instance of the notion of autism present in those diagnostic manuals. In this context, to say an autistic person knows autism best means they are more accurately able to establish they meet official diagnostic criteria than official diagnosticians. When self-diagnosing based upon DSM or ICD notions of autism, there is a diagnostic checklist which should be followed. It has quite specific criteria whereby an individual needs to exhibit multiple symptoms from one list (differences in social communication and social interaction) and multiple symptoms from a second list (restricted and repetitive behaviors). Consequently, an individual can exhibit multiple symptoms of autism but does not meet the DSM and ICD diagnostic criteria for autism. This means a self-diagnosing individual who accepts the DSM and ICD should consult the DSM and ICD diagnostic criteria for autism to see what the specific criteria are.

The diagnostic criteria are very easy to find through an internet search, so arguably the vast majority of self-diagnosing individuals have the means to check the DSM and ICD diagnostic criteria. I am not aware of any statistical information about how many do check the diagnostic criteria. Even if, as seems likely, the vast majority do check the diagnostic criteria, there is the issue of the impact on the accuracy of self-diagnosis if a small proportion do not check official diagnostic criteria. For self-diagnosis to be in line with the DSM and ICD to be accurate, the vast majority of self-diagnosing individuals need to check their diagnostic criteria, and any small minority who do not and must not result in a significant impact on the overall accuracy of self-diagnosis.

However, self-diagnosing individuals arguably need to do more than just check the official diagnostic criteria because those diagnostic criteria only convey limited information. The DSM specifies that the diagnostic criteria “are illustrative, not exhaustive; see [the rest of the] text” (APA, Citation2013, p. 50). The diagnostic criteria cover only one and a half pages of the 10-page entry on autism (APA, Citation2013, pp. 50–59), whereas the rest of the text, which is not readily available on the internet, covers eight and a half pages. As such, official diagnosticians with access to a physical copy of the DSM will have more detail on how the DSM portrays the diagnostic criteria than is available to the general public. More fundamentally, most official diagnosticians will have read far more about autism than the DSM or ICD. They will typically have at minimum a working knowledge of the scientific literature on autism. As such, their knowledge of the scientific literature will likely influence their diagnostic practice. Some of this may increase accuracy by providing accurate knowledge about autism, but some might instead decrease accuracy by reinforcing misleading stereotypes about autism.

Given the lack of details of the DSM and ICD diagnostic criteria, the accuracy of self-diagnosis arguably requires knowledge of autism, which extends beyond checking the diagnostic criteria. Four places where self-diagnosing individuals can gather potentially relevant information are scientific literature, websites, internet forums, and social media. It is possible to find an immense level of information about autism through open access or pre-print scientific literature, through blogs, through healthcare websites and charity websites, through internet forums dedicated to mental health or particular psychiatric diagnoses and through general social media as well as social media dedicated to mental health or particular diagnoses. It is worth noting that these sources can provide differing levels of accurate information and have differing levels of relevance. It is contestable as to which sources are more accurate and more relevant. For example, it might seem intuitive that journal articles published by trained scientists will contain more accurate descriptions of autism than blog posts written by people with no scientific training. However, it can be argued that scientists hold inaccurate stereotypes of autism while a blog post based upon lived experience can be more accurate despite the author having no scientific training. Establishing which sources are best might depend upon wider questions that I have already raised. For example, if introspection is taken as highly reliable, then a blogger with lived experience might be favored, whereas if introspection is highly unreliable, then perhaps scientific observations of external behavior might be favored. As such, I restrict myself to analyzing the consequences of someone who is self-diagnosing having good or bad information rather than specifying which sources provide good or bad information.

Self-diagnosing individuals will vary in the degree to which they access this information, and the quality of this information will also vary significantly. For those self-diagnosing individuals who do research psychiatric diagnoses through these means, there becomes the question of the degree to which they read and understand the type of information that increases diagnostic accuracy. Also, there is the question of the degree to which they do not read the type of information which would decrease accuracy, or they are able to recognize the problems in any of the flawed information they do read.

One potential way of avoiding this problem is by appealing to lived experience. It could be argued that an autistic person does not need to gain significant information about autism through scientific literature, websites, internet forums, and social media to understand autism because they can rely upon self-knowledge. Similarly, it can be argued that autistic individuals who do read flawed information from these sources can draw upon lived experience to recognize that it is flawed. However, this still presupposes that self-knowledge provides knowledge of autism. Using self-knowledge requires an interpretation. They need to take their self-knowledge and correctly interpret it as relating to autism. This is potentially problematic because someone could have a very good understanding of how they behave, perceive, think, and feel without accurately establishing which diagnosis they fit. The same behavior, perception, thought, and feeling can be associated with multiple diagnoses (Fellowes, Citation2022). For example, experimental and clinical evidence suggests that multiple diagnoses are associated with the simultaneous presence of both social communication differences and repetitive/restrictive behavior. Both of these are simultaneously present in autism (APA, Citation2013, p. 50), ADHD (Antshel & Russo, Citation2019, p. 3), schizotypal personality disorder (Hurst et al., Citation2007, 1712), and schizoid personality disorder (Sugihara et al., Citation2008, 1998) but they appear to manifest in subtlety different ways. As such, an individual might think they exhibit social communication differences in a manner associated with autism but actually exhibit them in a manner associated with ADHD, schizoid personality disorder, or schizotypal personality disorder. This suggests that an autistic person can only accurately interpret self-knowledge in relation to self-diagnosis through having an understanding of autism and other relevant diagnoses.

Drawing upon self-knowledge only starts being accurate for self-diagnosis when a level of information about psychiatric diagnoses is already in place. Once that level of information is present, then an autistic person can draw upon lived experience to interpret information about autism. Consequently, lived experience does not mean accurate information about autism is no longer needed. The advantage of lived experience means an autistic person would need to read less high-quality information about autism compared to a non-autistic person to generate a good understanding of autism. It also means an autistic person is better able to read low-quality information about autism without then generating a flawed understanding of autism compared to a non-autistic person. As such, appealing to lived experience only partially resolves the problem that accurate self-diagnosis requires information about autism. An autistic individual needs less information about autism than a non-autistic individual, but some is still needed.

This appeal to self-knowledge to mitigate the problem of limited or flawed information is harder to apply to non-autistic individuals. For self-diagnosis to have high accuracy it seems that non-autistic individuals need to be able to recognize that they are not autistic. However, for obvious reasons, individuals who are not autistic cannot draw upon lived experience of autism to realize that they are not autistic. Rather, non-autistic individuals will have lived experience of something else, such as being a typical member of the population. As such, they cannot appeal to lived experience of autism to reduce the level of high-quality information about autism they need to understand or appeal to lived experience of autism to help recognize the flaws in low-quality information about autism. The lived experience of a typical member of the public may result in a nuanced understanding of themselves, but it is unclear how this could then result in a subtle understanding of autism. As such, if lived experience of autism means that autistic individuals can accurately self-diagnose as autistic, it is unclear if lived experience can play a similar role when non-autistic individuals assess if they are autistic. This would mean self-knowledge would decrease false negatives, reducing the number of autistic people failing to judge that they are autistic but would not prevent false positives, not reducing the number of non-autistic individuals incorrectly judging that they are not autistic. Either an argument is needed that non-autistic individuals can use their self-knowledge of non-autism to understand the subtle nature of autism or some other means of understanding the subtle nature of autism is required. Failing this, the accuracy of self-diagnosis is reduced.

To summarize, when self-diagnosing whilst accepting the DSM and ICD, an individual needs to avoid the problems of introspection and cognitive biases. They also require knowledge of the DSM and ICD notions of autism, and this typically requires going beyond the diagnostic criteria themselves. They need to adequately understand good information about autism whilst not being influenced by bad information. Someone who actually is autistic can potentially and partially reduce the required level of good information and can more easily spot bad information, whereas this resource does not appear open to someone who is not autistic. Unless a means of mitigating these issues is established, the accuracy of self-diagnosis is reduced.

9. Self-diagnosing when rejecting the DSM and ICD

I now focus on self-diagnosing individuals who reject official diagnostic criteria. The key idea of self-diagnosing when rejecting official diagnostic criteria is that the self-diagnosing individual has better knowledge than is conveyed by the official diagnostic criteria. As such, the debate is not over whether the self-diagnosing individual meets the official diagnostic criteria. Instead, the debate is about what the diagnostic criteria should be. The claim that a self-diagnosing person knows something which is not reflected in the diagnostic criteria works differently depending upon what psychiatric diagnoses are taken to be. I will outline three possibilities, namely psychiatric diagnoses as objective entities, as the product of theoretical virtues and as the product of practical values.

10. Conclusion

I have considered what arguments are needed to show that self-diagnosis is accurate. I have identified four different approaches to claiming that self-diagnosis is accurate. These are, firstly, accurately assessing whether they meet the diagnostic criteria in the DSM and ICD, accurately assessing whether they are an instance of an objective entity, accurately formulating a diagnosis that is theoretically virtuous, and accurately formulating a diagnosis that realizes practical factors. Some of the required arguments to justify the accuracy of self-diagnosis are applicable to all the different approaches, namely those arguments relating to introspection and cognitive biases. However, all the approaches also need some unique arguments which are not present in other approaches. I hope this article has contributed through identifying different approaches to justifying the accuracy of self-diagnosis, and by showing that different arguments are needed for different approaches. I hope this will be a stepping stone toward establishing whether those arguments, and which of those arguments, can be successfully justified.

If self-diagnosis is inaccurate, then the question whether legitimate self-diagnosis requires accuracy becomes important. Relevant issues here are establishing the exact degree of inaccuracy, whether only some types of self-diagnosis are inaccurate, whether the accuracy of self-diagnosis is lower than the accuracy of official diagnosis, whether individuals mis-diagnosing themselves as autistic is harmful to themselves and whether individuals mis-diagnosing themselves as autistic is harmful to others. I hope this paper will inspire further work on these questions.