ABM atypicalities in autism have been noted since the earliest clinical reports of the condition. Kanner (1943) documented how children with autism failed to voluntarily recall events that happened during their day, and the parents of these children reported how sharing experiences “seemed foreign” to them. He also noted that when these children were questioned about their experiences, their responses were often incor- rect. In contrast, Asperger (1944, 1991) reported that children with autism did occa- sionally relay personal memories, but that these tended to be recalled in an odd or atypical context; for example, in Frith’s (1991) annotation of Asperger’s original paper “Autistic Psychopathy in Childhood,” she commented on how responses on the “similarities” subtest of a standard intelligence scale (in which participants were asked to note how two words were similar) yielded a “seamless” mix of general knowl- edge and autobiographical experience. In contrast, typical children would be able to
realize that an appropriate answer would involve providing a general knowledge response, rather than a specific ABM. Asperger (1944) also noted that when these children recalled events (usually in minute detail), they tended to forget information and to confuse details of the events. These early clinical reports of poor ABM were subsequently confirmed in empirical studies on autistic children.
In relation to autistic adults, findings from experimental studies (reviewed later in this chapter) appear at odds with the remarkably detailed and vivid accounts of past experiences documented in autobiographical writings. For example, Temple Grandin (2006, p. 102) describes in great detail some of her early school memories:
I vividly remember learning about the solar system by drawing it on the bulletin board and taking field trips to the science museum. Going to the science museum and doing experiments in my third‐ and fourth‐grade classrooms made science real to me. The concept of barometric pressure was easy to understand after we made barometers out of milk bottles, rubber sheeting, and drinking straws. We taped the straw onto the rubber sheeting, which covered the mouth of the milk bottle. Changes in the air pressure pushed the rubber membrane up and down and made the straw move.
A further paradox, as noted earlier, regards the retrieval of early ABMs being an area of relative strength in autism. While the literature on “childhood amnesia” sug- gests that typically developing individuals tend to recall their earliest memories from around the age of three and a half (Pillemer & White, 1989), Lyons and Fitzgerald (2005) noted that a number of individuals with autism possess a very early memory, from around 1 year of age (Lyons & Fitzgerald, 2005). This may relate to the utiliza- tion of different ABM organizational strategies in adults with autism (with typical adults relying more on pre‐existing schemas than autistic adults). This is, however, specula- tive; this area is vastly under researched and requires rigorous and systematic investi- gation in future. Nevertheless, the question remains of how we account for these seemingly contradictory findings from experimental studies alongside this anecdotal data. Next, we review the experimental literature.
Episodic versus semantic autobiographical memory in autism
One of the first published group studies of ABM in adults with autism was by Goddard et al. (2007). In this study, the authors examined the recall of personally experienced events in adults with and without autism. This was achieved using the ABM cueing task (Williams & Broadbent, 1986), whereby participants were presented with a cue (in this case, a word) and were instructed to generate a specific ABM—a memory of a particular event, occurring in a particular place and time, lasting no longer than a day (Conway & Rubin, 1986)—at speed. There were two outcome measures on this task: (a) the number of specific memories retrieved to cue words, and (b) the time taken to retrieve a specific memory. Using positive, negative, and neutral word cues, Goddard et al. (2007) found that autistic adults recalled significantly fewer specific ABMs, and took considerably longer to do so, than a comparison group matched for age, gender and IQ.
Goddard et al.’s (2007) study provided three other important insights into the nature of ABM in this group. First, this study revealed that the ABM difficulties displayed by the autism group were independent of challenges with social‐problem solving.
In typical adults, ABM and social‐problem solving are closely linked, with past experi- ences being used to guide future behaviors. For example, when arguing with a friend, one may reflect on previous arguments (and what conflict resolution strategies worked well/did not work well) in order to decide on an appropriate course of action on this current occasion. In line with this, Goddard et al. found a correlation between ABM and social‐problem solving in their sample of typical adults. In contrast, they found no correlation between ABM and social‐problem solving abilities in adults with autism.
This suggests that autistic adults may find it difficult to use their memories in an adap- tive way (to guide future behaviors). For example, they may repeatedly make the same errors when resolving conflicts, without learning from their past experiences.
Second, Goddard et al. (2007) reported that the specific ABM difficulties displayed by the adults with autism appeared to be largely independent of a general memory deficit. However, visual memory difficulties were observed in their sample of autistic adults, and these were significantly (negatively) correlated with ABM performance.
Indeed, ABMs are thought to be stored visually (Brewer, 1986) and this may be one potential mechanism underlying the ABM difficulties of autistic adults.
Third, Goddard et al.’s study provided insight into the role of emotion in ABM.
Whereas the typical comparison group used emotion cue words (both positive and negative) to facilitate specific ABM retrieval, this pattern was not observed in the autism group (there were no differences between the number of memories retrieved or the latency to memory retrieval, depending on whether cues were positive, nega- tive, or neutral). This finding does need to be treated with caution, however, as it was not replicated in a later study by Crane, Goddard, and Pring (2013). The role of emo- tion in ABM is discussed again later in this chapter.
Although Goddard et al.’s (2007) study was the first published group study to demonstrate ABM difficulties in autistic adults, only one aspect of ABM was explored—memories of specific autobiographical events. As previously noted, ABM comprises both personally experienced events (episodic ABM) and self‐related infor- mation (semantic ABM), which combine to form a single ABM system. In an attempt to investigate both of these facets of ABM in autism, Klein, Chan, and Loftus (1999) reported the single case study of R.J., a 21‐year‐old man diagnosed with autism. R.J.
had accurate knowledge of his personality traits (his semantic ABM) in the absence of access to the specific events on which this knowledge was based (his episodic ABM), suggesting a dissociation between episodic and semantic ABM in autism.
However, as these conclusions were based on the results of a single case study, it was unclear whether the results generalized to individuals with autism more broadly (especially considering the wide heterogeneity characteristic of the autism spectrum).
In addition, the researchers compared R.J.’s performance against that of a mental age‐matched comparison group comprising three 12‐year‐old children. Therefore, the size and quality of the memory databases of participants was likely to be substan- tially different.
To overcome these methodological issues, Crane and Goddard (2008) examined episodic and semantic ABM in a group of 15 autistic adults and 15 comparison adults matched for age, gender, and IQ. Using ABM tasks indexing both episodic and semantic ABM (two semi‐structured interview tasks and an autobiographical fluency task), the authors found evidence of episodic ABM difficulties in the autism group, coupled with preserved semantic ABM, a pattern consistent with that reported by Klein et al. (1999). Of interest, Crane and Goddard (2008) explored episodic and
semantic memories across different lifetime periods and reported that although the adolescent and early adult lifetime periods facilitated memory retrieval in the typical adults (known as the “reminiscence bump”), this was not the case for the autism group. This therefore suggests that this group may not be using the self to facilitate memory in the same way as typical adults (Conway & Pleydell‐Pearce, 2000).
Following on from this finding, the next section further explores the relation between the self and ABM in autism.
The self and ABM in autism
Using predictions derived from Conway and Pleydell‐Pearce’s (2000) model of ABM and the self‐memory system, Crane, Goddard, and Pring (2009) tested the sugges- tion that what we remember is strongly influenced by the goals of the “working self”:
a component of Conway and Pleydell‐Pearce’s (2000) self‐memory system, which modulates access to autobiographical knowledge by successively shaping retrieval cues (via executive processes). These retrieval cues are then used to activate autobiographi- cal knowledge structures, leading to the emergence of ABMs.
Using a paradigm adapted from Moberly and MacLeod (2006), Crane and col- leagues (2009) asked autistic and non-autistic participants to select goals that they were currently pursuing (from a list of 50 items). For each goal selected, participants com- pleted a questionnaire to determine whether the goal was self‐concordant (i.e., they were pursuing the goal because they wanted to) or nonself‐concordant (i.e., they were pursuing the goal because they had to). A subsample of three self‐concordant, three nonself‐concordant and three (random) nongoal items were then selected for each participant. First, a task indexing general ABMs was administered. Here, participants were shown (fairly quickly, in under 4000 ms) a series of short (two‐word) phrases on a computer screen (e.g., romantic experiences). The participants had to respond (by pressing a Yes or No key) as to whether they had generally experienced that event in their own life, making their decision as quickly as possible, but ensuring that their response was considered. This task did not require participants to access a specific memory prior to making a decision (indeed, it was emphasized that there was no need to do this); rather, this task was purely related to their rough impressions of whether or not they had generally experienced the event. Memories of specific autobiographical events (referred to in Conway & Pleydell‐Pearce’s model as “event‐specific knowl- edge”) were also assessed, using a cue word task adapted from Moberly and MacLeod (2006). Here, participants were presented with three self‐concordant and three non- self‐concordant goal‐related items, as well as three nongoal items, from which to retrieve memories of a single event, lasting no longer than a day (i.e., a specific ABM).
For the general memory task, Crane et al. (2009) found that the autism group took longer to respond, but both groups showed a similar pattern of responding as a function of the memory cue: General event knowledge accessed in relation to self‐
concordant goals was accessed quicker than that accessed to nonself‐concordant goals, which was accessed quicker than knowledge retrieved to nongoals. However, a very different pattern appeared for event specific knowledge. For typical adults, spe- cific memories retrieved in relation to goal cues were accessed faster than those retrieved in relation to nongoal cues. Goal self‐concordance did not play a role, probably due to the crudeness of the measure of memory latency (timed via a stopwatch, rather than in ms). In contrast, the autistic group did not show this pattern. Of importance,
this did not appear to be due to task demands. Administering a reverse version of the ABM cueing task (in which participants were given a cue and asked to respond with a general, as opposed to a specific, memory), Crane et al. (2009) found that adults with and without autism could use goal cues to facilitate general memory retrieval. Overall, these studies suggest that the general, but not specific, ABMs of autistic adults appear to be similar to those of typically developing individuals. Further, they suggest that the specific ABM difficulties of autistic adults may be related to problems in using the self as an effective memory organizational system. This may be compounded by the greater executive resources inherent in specific (opposed to general) ABM retrieval (Conway & Pleydell‐Pearce, 2000), particularly given that executive functioning dif- ficulties are well established in autism (Hill, 2004).
Crane, Goddard, and Pring (2010) further explored the role of the self and ABM by assessing “self‐defining memories” in adults with autism. As previously noted, self‐defining memories are a specific component of ABM that define our sense of self.
In this study, Crane and colleagues (2010) found that, contrary to predictions, autis- tic adults were able to generate self‐defining memories and the quality of these was similar to those recalled by typical adults. However, their self‐defining memories were not as specific as those recalled by typical adults (they were less likely to be focused on one specific day). An exploration of memory narratives provided one possible expla- nation for this: Whereas the narratives of the typical adults commonly referred to events concerning achievement/mastery (to a greater extent than those of the autistic group), the memories of the autistic adults centered more heavily on recreation/
exploration (consistent with the restricted and repetitive interests that form part of the autism diagnostic criteria; American Psychiatric Association, 2013). While recrea- tion/exploration narratives were largely general in nature (e.g., “Skiing has always been a big part of my life”), achievement/mastery events often centered on one spe- cific day (e.g., “when I received my exam results”).
Perhaps a more important difference between the narratives of the autistic and com- parison groups regards how the autistic adults extracted significantly less meaning from their narratives than their typical peers. This is illustrated in the following examples taken from the study by Crane et al. (2010). In this first example, the participant describes a situation involving one of her favorite interests (a band). While this autistic adult reports something quite out of character, she does not report learning any lessons from the event, nor does she explain how it may affect any of her future behaviors.
Example one: autistic adult. My favourite band did their first show for something like fifteen years and it was the first time I’d ever seen them on tour and I was standing very near the front but when they came on, suddenly this crowd pushed in and shoved me right the way through and I was just jammed in and it was actually swelteringly hot, so much so that the security guards were handing out water to the audience and were get- ting quite concerned about the audience, some people fainted and things like that. And I kind of managed to enjoy it anyway, even though the conditions were absolutely hor- rific. I mean it’s very out of character for me to enjoy big crowds and touch, it’s just what I was prepared to do just to see this band.
In contrast, the non‐autistic participant in the next excerpt describes the death of a relative and goes on to discuss the impact that it has had on her and how she has learned important lessons from the experience that will guide her future behaviors (meaning making).
Example two: typical adult. Basically my uncle died of alcoholism last year in October, so that was about a year ago, the whole family were obviously devastated by the event, but the thing that I actually learnt from it was not to actually get addicted to a substance that could actually control me and to monitor the amount I drink, and also not to drink and drive and to seek help if I ever did wish to go down that avenue and it’s also bought my family closer together because now we’re kind of more open about how we really feel so no one will ever feel the need to shut themselves out to the world and let something like alcohol control them.
This finding of less “meaning” being derived from the narratives of autistic individu- als is consistent with two theories of cognition in autism. First, the Enhanced Perceptual Functioning model of Mottron and colleagues (Mottron & Burack, 2001; Mottron, Dawson, Soulieres, Hubert, & Burack, 2006) stresses that the relation between high and low level cognitive processes is atypical in autism, with the superiority of perceptual processing in autism being disruptive to other higher level cognitive abilities. This reli- ance on lower level perceptual processing may result in a reduced tendency to extract high‐order meaning from narratives. Second, Bowler and colleagues (Bowler & Gaigg, 2008; Gaigg, Gardiner, & Bowler, 2007) propose that relational memory (the ability to identify meaningful links between to‐be‐remembered items that is required for meaning making) is an area of difficulty in this group (see Chapter 1 for a review).
Nevertheless, one possibility that cannot be discounted is that the autism group were able to extract meaning from their memories, but opted not to report instances of lesson learning or gaining insight within their narratives. As stressed by Crane and colleagues (2010) in their original paper on this topic, the task instructions only asked participants to describe the memory—there was no explicit requirement to report instances of meaning making. This accords with White, Burgess, and Hill’s (2009) suggestion that the difficulties faced by autistic people on open‐ended cognitive tasks (i.e., tasks in which there are various potential approaches to responding, for which participants themselves must decide is the most appropriate course of action) may be due to a lack of “implicit mentalizing”—the ability to appreciate the inherent require- ments of a task and respond to the social expectations of the experimenter. It is therefore important for future research to explore whether autistic adults can identify meaning from their narratives (by explicitly asking them to do so), and whether they can use this information in a functional and directive way.
Findings of a largely unimpaired “self‐enactment effect” (whereby information processed in relation to the self is better remembered than information relating to, or actions performed by, another person) indicate that self‐related information is pro- cessed at a deeper level by autistic individuals (e.g., Grainger, Williams, & Lind, 2014;
Maras, Memon, Lambrechts, & Bowler, 2013; Williams & Happé, 2009; and see also Chapter 4; Lind, 2010; and Lind, Williams, Bowler, & Peel, 2014). Nevertheless, it is important to note that the atypical specific ABM relations with the self in autism (such as links to goals and extracting meaning from very salient narratives) are likely to be more sophisticated than the processes that underlie the self‐enactment effect.
Although some researchers have viewed episodic memories and ABMs as mutually exclusive, ABMs are thought to be much richer in nature, involving a greater degree of self‐relevance and requiring the integration of events within one’s personal past (Conway, 1990). Thus, an intact self‐enactment effect may not simply equate to undi- minished self‐related ABMs, and further research is needed to disentangle the com- plex relation between the self and specific ABM in autism further.