The categorisation of resistance: interpreting failure to follow a proposed line of action in the diagnosis of autism amongst young adults.

Many characteristics typical of autism, a neurodevelopmental condition characterised by socio-communicative impairments, are most evident during social interaction. Accordingly, procedures such as the Autism Diagnosis Observation Schedule (ADOS) are interactive and intended to elicit interactional impairments: a diagnosis of autism is given if interactional difficulties are attributed as a persistent quality of the individual undergoing diagnosis. This task is difficult, first, because behaviours can be interpreted in various ways and, second, because conversation breakdown may indicate a disengagement with, or resistance to, a line of conversation. Drawing upon conversation analysis, we examine seven ADOS diagnosis sessions and ask how diagnosticians distinguish between interactional resistance as, on the one hand, a diagnostic indicator and, on the other, as a reasonable choice from a range of possible responses. We find evidence of various forms of resistance during ADOS sessions, but it is a resistance to a line of conversational action that is often determined to be indicative of autism. However, and as we show, this attribution of resistance can be ambiguous. We conclude by arguing for reflexive practice during any diagnosis where talk is the problem, and for a commitment to acknowledge the potential impact of diagnostic procedures themselves upon results.


37
In recent years, there has been an increasing focus in the academic literature on communication 38 between those with autism, and those they interact with. Autism is defined as a 4 offer alternatives, therefore, this is still a form of 'controlled autonomy' (Callon & Rabeharisoa 77 2002: 13) where engagement ensures that some possibilities arise while others are eliminated 78 (Hacking 1995: 241;Hollin 2017: 617). Second, acts of resistance may be co-opted by those being 79 resisted. So, for example, it has been argued that self-advocacy and the social model of disability,  Interactional resistance and autism 91 As we noted at the outset, core symptom clusters in autism manifest themselves most clearly 92 during interaction and conversation. Diagnostic procedures such as the Autism Diagnosis 93 Observation Schedule (ADOS: , which we will consider in more detail in the 94 following section, often involve structured interaction and are intended to elicit such symptoms. 95 A diagnosis of autism is given if interactional difficulties, which diagnosticians are attempting to 96 elicit, are attributed as a persistent quality of the individual undergoing diagnosis 1 . However, and In this paper, we focus on the production of interactional resistance during the actual process of 122 testing for diagnosis. As we have described above, resistance to the ultimately proferred diagnosis, 123 or to the visible building towards such a diagnosis (see , is a previously observed 124 feature of healthcare interactions, and is generally treated as an understandable response to 125 unexpected or unwelcome news. Furthermore, previous conversation analytic work begins to 126 identify the range of interactional forms resistance may take. It may be expressed directly, as in the 127 case of the parents explicitly resisting a label of intellectual disabilities for their child (Gill and 128 Maynard 1995), or it may be more indirect or passive, as Heritage and Sefi demonstrate in response 129 to unsolicited advice from Health Visitors and as Stivers (2007) identifies in relation to parents' 130 treatment of doctors' refusal to prescribe desired antibiotics to their children because of a viral 131 diagnosis. The question that guides our analysis here is: how, in the case of autism, do 132 diagnosticians distinguish between interactional resistance as, on the one hand, a diagnostic 133 indicator and, on the other, as a relevant and reasonable choice from a range of possible responses? 134 Put simply, how is it to be judged whether resistance in this setting is to be considered 'mundane' 135 or 'autistic'? 136 It is important to note here that it is not our contention that autism is purely a social construction 137 (see similar arguments made in relation to intellectual disability, e.g. Rapley (2004)). We do not 138 advocate for institutional determinism, deny the reality of individuals' difficulties with social 139 interaction, seek to undermine the judgement and effort made by diagnosticians or the fact that a 140 diagnosis can provide individuals and families with much needed access to support and resources.

141
Our focus is purely on a specific empirical problem which must be resolved interactionally: by 142 what criteria can everyday interactional practices be distinguished from diagnostic indicators? It is 143 also important to note that we seek to assign no blame, or pass any judgement, on the professionals 144 whose interactions are presented here. We simply seek to shed light on the different ways resistance 145 may manifest itself in these interactions, and the practical problem of categorisation that then 146 7 arises. In order to unpack this, we focus on the actual delivery of the ADOS test and how it is 147 assessed in the moment, through interactional interpretative work.

148
The ADOS test 149 Diagnostic scoring and rating instruments are commonly used across a wide range of medical 150 specialties, from the APGAR scores applied to newborn infants to the Mini Mental State 151 Examination (MMSE) used in the diagnosis of dementia. By their nature, such instruments pre-152 define specific issues or behaviours as significant. However, as Turowetz has noted, the 153 contribution of these instruments to the production of diagnostic 'facts' is generally minimized; 154 they are 'treated as neutral, autonomous tools of measurement that record data for assessment, 155 rather than contributing to such data' (Turowetz 2015a: 215 The Autism Diagnostic Observation Schedule (ADOS) was first developed in 1989 (Lord et al. 163 1989), intended for use within both research and clinical settings (Lord et al. 1989: 186), and for 164 those with a verbal age greater than three (Lord et al. 1989: 208). Subsequently, and with the intent 165 of facilitating clinical evaluation in a wider range of individuals, the 'ADOS-generic' (ADOS-G) 166 was developed in 2000 . This new version of the ADOS has four sub-versions 167 ranging from a 'module 1' version intended for preverbal individuals through to a 'module 4' 168 version intended for adolescents and adults with fluent speech.

169
In all its forms the ADOS is:  : 187, italics in original) 173 In the social psychological tradition, therefore, the highly trained practitioners giving the ADOS 174 are understood as stooges or confederates (Lord et al. 1989: 187), standardising activities and their 175 own behaviours in order to prompt a number of 'social occasions' within which 'a range of social 176 initiations and responses is likely to appear' (Lord et al. 2000: 205). These invitations are referred 177 to as 'presses'. Presses on module 4 for the ADOS include: engaging in conversation about a range 178 of 'socioemotional' issues (e.g. friends, loneliness, social difficulties) and everyday functioning 179 (school/work); a construction task (akin to making a simple jigsaw); telling a story from a picture 180 book; physical demonstration of an everyday task (e.g. brushing of teeth); creating a story with the 181 use of physical objects (including, in our sample, a toy car, a sponge, and a cocktail umbrella); the 182 retelling of a cartoon strip; free play with toys; and description of a picture featuring a social scene.

183
Throughout these activities the investigator searches for the social and communicative atypicalities 184 associated with autism. While accounting for the possibility of resistance during these activities 185 does not appear to have been a priority to the creators of the ADOS, measures were put in place 186 to address expected ambiguities. The 0-3 rating scales described below are intended to allow 'room 187 for uncertainty' (Lord et al. 1989: 190)   Following the ADOS sessions, which are video recorded, examiners watch back the video -191 sometimes though not always with a colleague -and score participants' behaviour across a range 192 of domains. In some areas the scoring criteria frame this as a quantitative exercise; 'imagination 193 and creativity', for example, is scored from 0 (several instances where imagination is demonstrated) 194 to 3 (no instances where imagination is demonstrated). In other areas, examiners are required to 195 make a more explicitly qualitative assessment; 'overall rapport' for example is ranked between a 196 'comfortable' 0 and an 'uncomfortable' 3. Upon conclusion, participant scores are added and a 197 diagnosis of 'autism' is given for particularly high scorers, 'autism spectrum' for those scoring 198 reasonably highly, or 'non-spectrum' for low scorers. While the ADOS has well recognised clinical 199 limitations which prevent its use in isolation -for example, it was designed neither to examine age The current sample 205 The current study examines 7 ADOS sessions, all of which were conducted using the 'module 4' 206 version of the test . The individuals undertaking the ADOS were all men and 207 aged from late teens to mid-twenties. All had pre-existing diagnoses of either Asperger's Syndrome 208 or autism and, on the basis of these diagnoses, had been invited to take part in a university-based 209 research study for which it was necessary that a further ADOS be completed. In every case a 210 diagnosis of autism or autism spectrum was confirmed. The two examiners conducting the ADOS 211 were both female postgraduate students in their twenties and had been fully trained and qualified 212 to administer the procedure (Lord et al 2002). Although only one examiner acted as 'stooge' in any 213 given session both were involved in the rating of all participants.

214
In the light of the above information it should be noted that there is a particular dynamic within 215 this sample. Both participants and examiners already knew that an independent diagnosis of 216 autism/autism spectrum had been arrived at previously and there were no clinical consequences 217 following the current sitting (i.e. existing diagnoses could not be questioned). Such uses of the 218 ADOS are intended : 186) but, as we stress in the analysis and discussion, 219 generalisations to other contexts should be made with caution. 220 10 ADOS sessions took place in either the participant's educational setting or at the researchers' 221 university and lasted between 35 and 52 minutes 2 . As is typical (see above) these sessions were 222 recorded in order to facilitate scoring and it is these videos -and the note and scoring sheets made 223 by the examiners -which are utilized in the present study. The note sheets were taken by the 224 examiners either during the ADOS sessions or immediately afterwards. The score sheets contain 225 not only the final diagnostic judgements but also the 'working out' of these scores (so for example 226 noting how many and where instances of 'demonstration of imagination' occurred in order to 227 assign a number from the scale). These written documents therefore provided a significant insight 228 into the diagnostic production process. The present study received ethical approval following 229 University ethical review procedures and all participants gave written permission for their data to 230 be reused for this piece.

232
The video-recorded ADOS sessions were fully transcribed using CLAN software, and analysed recording what has been taken to be an example of a particular phenomenon), this meant that, as 245 far as possible, we could analyse the talk alongside the coding categories that had been assigned to 246 it.

248
Our analysis identifies three different kinds of resistance in our data, which will be considered in 249 turn: resistance to a proposed task; resistance to a behaviour or feeling being characterised in a 250 particular way; and resistance to a proposed line of conversational action. 251 1) Resistance to a proposed task. 252 In our data, this kind of resistance is produced in response to requests to participate in specific 253 components of the test, for example a request to act out an action, and a rationale is usually 254 provided for the refusal (e.g. 'not with him watching'). In everyday interaction, resistance to comply 255 with requests is dispreferred, with CA research repeatedly demonstrating that human interaction 256 is organized to favour actions promoting social affiliation (Pillet-Shore forthcoming; Pomerantz 257 and Heritage 2012; Kitzinger and Frith 1999). As a result, a refusal is usually produced with an 258 account or a mitigation; where it is not, it may be seen as accountable by the requesting party 259 (particularly where the request is produced with a high degree of entitlement and a lack of 260 contingency (Curl and Drew 2008)) and so be pursued by the requestor. This pursuit commonly 261 takes the form of reframing. Reframing generally treats the resistance as either a lack of 262 understanding (so the requestor goes on to describe it differently) or a lack of ability (e.g. that 263 someone can't reach something they've been asked to pass). It is not generally treated by the co- In this extract, the interviewer introduces a new component of the ADOS test, that of acting out 286 an everyday action. However, she prefaces her description with an acknowledgement that the 287 request which will follow is potentially problematic, and may make the participant 'feel a bit silly' 288 (line 581). While the interactionally preferred response to a request is acceptance or compliance, 289 this request, then, is designed in such a way as to make refusal easier; itself a demonstration that Extract 2 below shows a further example of resistance to a proposed task, when the interviewer 304 invites the participant to tell a story using the objects on the table; in this instance these include 305 small toys such as a car and a ball, and small household items such as a shoelace and a cocktail 306 umbrella. Immediately prior to this extract, the interviewer has explained the task by telling a short 307 story using these objects herself. where this phenomenon tends to be interpreted as part and parcel of the underlying condition. It

347
appears that what the kind of resistance displayed here does, perhaps paradoxically, is to enable 348 participants to demonstrate interactional competency. We will now turn to examine the second 349 category of resistance emerging from these data. phenomenon as a justification for arguing that more directive styles of interaction may therefore 400 be appropriate. However, in this instance, the long term effect of presenting something which is 401 seen as an 'abnormal' or 'accountable' way of dealing with feelings of sadness is both anticipated 402 and set aside by this young man's response.

403
In the second example of this type shown here, the interviewer has just concluded the telling of 404 her story using the everyday objects that are available (this process is detailed in the discussion of 405 extract 2). Before she began, she informed the participant that after she had completed her story, 406 she would ask him to tell one. At the conclusion of her story there is shared laughter, before she 407 invites the participant to begin by way of referring to the quality of her own story: then the resistance is more subtle than in Extract 3, and is only temporary.

428
In both of the extracts shown here, the resistance which is displayed moves beyond the interaction.

429
In so doing, it orients to the fact that this is not simply a conversation where regular conversational 430 actions have to be attended to (responding to a question, listening to a story etc) but also one 431 where both the quality and content of these actions are being assessed through a particular 432 framework. As with the first type of resistance identified above, in our data this type of resistance 433 is likely to be treated as a reasonable, 'normal' response. In neither of the examples above do 434 testers score this resistance as problematic. 435 We now turn to examine our third category of resistance: resistance to a line of conversational 436 action. As analysis will show, this category is both more complex and more consequential than 437 those considered previously. In this instance, the interviewer already knows that the respondent is a physics graduate, and they 494 have been talking about a third party who has recently completed their Masters degree. The

495
interviewer makes a proposal as to the kind of qualification in physics the respondent has, which 496 he affirms in line 372. The interviewer responds by relating that this is the same qualification as 497 her fiancé has, but stating that 'he didn't do it in physics'. There is a lengthy pause, during which 498 there is eye contact and a small head movement in acknowledgement by the participant. However, 499 no verbal response is produced to either acknowledge the story or to seek further details. Again, 500 this failure to respond is categorised as problematic, with the notes identifying a failure to follow 501 up the interactional 'press'.

502
These kinds of 'presses' are included in the ADOS precisely because a failure to respond 503 appropriately to them is seen as characteristic of autism. An inability to engage in social 504 communication and a lack of awareness of another's feelings or emotions are diagnostic criteria of 505 autism, and so the failure to respond to presses like these is taken as the interactional manifestation 506 of autism. The diagnostic importance of these instances for the ADOS means that it is critical that 507 they can be accurately and appropriately identified. However, given the messiness of talk-in-508 interaction in general, we suggest that this task may be more challenging than is generally 509 acknowledged. Extract 7 below shows another example of a 'press' which is not responded to, 510 which we argue is much more ambiguous than the two we have seen so far: The notes suggest that this introduction of the interviewer's story about a rollercoaster was 549 intended as a press, and the lack of response is rated by one of the two scorers as problematic for 550 that reason. However, we would argue that a judgement about whether this instance is to be 551 considered consequential is very delicate, and complicated by the fact that there is an arguably Analysing the scoring notes alongside the three categories of resistance we have identified in these 588 data shows that it is this third category of resistance, resistance to a line of conversational action, 589 that is the most consequential in terms of its likely diagnostic implications. It appears, however, 590 that this is also the most interactionally complex category, so that the potential for ambiguity of 591 interpretation is greater. We argue that this ambiguity is related to the fact that any resistance 592 displayed in response to an ADOS interactional 'press' is inherently likely to be much more indirect 593 than that displayed in response to a request for action, or as a pre-emptive strike against 594 categorisation. Direct requests, for example, fit an adjacency pair format and conditional relevance 595 of the response is a normative requirement, so that pursuit of a request is both expected and 596 accepted where a relevant response does not occur. A failure to align as a troubles recipient in 597 response to someone else's expressed difficulties is both less straightforwardly accountable and 598 more likely to be done indirectly.

599
Previous CA work suggests that one way in which resistance may be indirectly expressed in 600 healthcare is by clients withholding a response to an expressed perspective ( given the degree of interpretation that is required.

26
It is however important to note that our sample is both small and particular. Participants were 630 intellectually able and demonstrated a degree of insight that may not be common across the 631 population with whom the ADOS is used. Participants also had a pre-existing diagnosis of 632 autism/autism spectrum and no clinical consequences followed from this particular encounter. As 633 we stress in the analysis, these factors limit the extent to which generalisation to clinical settings or 634 other populations is appropriate. Also noteworthy is the significant contrast between some of the 635 standard ADOS processes and the general norms of wider healthcare interaction; for example, it shows the difficulties clients experience where a non-directive ethos means that they are expected 641 to set their own agendas and understand that practitioners will not make testing recommendations 642 for them (Pilnick 2002a(Pilnick , 2002b.