In the past decades, autism researchers have begun to investigate strengths in the cognitive profile of individuals with Autism Spectrum Disorder (ASD). This may have been inspired by the seminal work by Shah and Frith (1983, 1993), who showed superior performance in autistic compared to typically developing (TD) children on the block design sub-components of the Wechsler intelligence scales (Wechsler, 1981) and the Children’s Embedded-Figures Task (Witkin et al., 1971). The authors noted that a key characteristic of both tasks was that the detail-focused and “local” performance benefit in ASD might have been based on a disruption of Gestalt grouping. As a result, the salience of parts (local processing) over wholes (global processing) might increase as distraction from the overall Gestalt is reduced. This could have caused an advantage for the autistic over the TD group (e.g., Falter et al., 2008), which is in accordance with the “Weak-Central-Coherence” theory (WCC, Frith, 1989) predicting superior local processing (Happé and Frith, 2006) and originally also reduced global processing (Frith, 1989). The assumption that autistic persons have a preference for local processing has indeed been widely supported (Mottron et al., 1999; Behrmann et al., 2006; Wang et al., 2007). Supporting this view, the results of a meta-analysis that included 35 studies comparing among others performance on Block Design and Figure-Disembedding found that autistic participants are, on average, superior at tasks that require detail-focused processing compared to control persons (Muth et al., 2014). On the other hand, the second (original) assumption of the WCC theory, concerning the corresponding global processing impairments, has been challenged (Brosnan et al., 2004; Wang et al., 2007; Scherf et al., 2008; Koldewyn et al., 2013; Hadad and Ziv, 2015; Booth and Happé, 2018). Indeed, some studies report that autistic individuals can successfully re-organize the integral structure of visual information. For example, researchers have found superior performance of autistic individuals on tasks such as conjunctive visual search, where a target needs to be identified among distractors (Plaisted et al., 1998). Moreover, Dakin and Frith’s (2005) review article provides evidence only for superior local processing in persons with ASD, but not for impaired global processing. In response to this, alternative accounts of the WCC have been proposed, such as the Enhanced Perceptual Functioning account (EPF, Mottron and Burack, 2001; Mottron et al., 2006) and the Reduced Generalization Account (RG, Plaisted, 2001). According to EPF, the local preference of autistic individuals is not a consequence of a global deficit but merely of superior perceptual processes. It further suggests that local preference can be overcome by global requirements, and that the ability to process globally is not restricted. Moreover, the “low-level” perceptual processes are characterized by an atypical autonomy with reduced influence by “high-level” processes (Mottron and Burack, 2001; Mottron et al., 2006). The RG account assumes that autistic individuals tend to differentiate more, but generalize less strongly between stimuli, leading to a more detail-focused processing style. In light of the evidence, the original WCC theory was honed to “detail-focused cognitive style” (Happé and Frith, 2006).

In summary, the preference for a local processing mode (if not otherwise instructed) by autistic individuals stands in contrast to that by TD individuals, who tend to show a global processing preference (if not otherwise instructed). This raises the intriguing question of whether the respective default processing style might be influenced dynamically under certain task designs. For instance, it has been shown that participants with ASD can process globally if explicitly instructed to do so, e.g., in a selective attention task (Plaisted et al., 1999; Mottron et al., 2003).

Priming is a traditional technique to modify individual differences in perception despite identical stimulus material. It is used to analyze the influence of situational variables on perception, cognition, feelings and behavior (Bargh and Chartrand, 2000). Indeed, a whole series of studies has demonstrated the effect of priming on the visuo-spatial perception of TD participants (e.g., Lin et al., 2008). Many researchers are of the opinion that situational factors in daily life trigger similar priming effects and therefore have a significant influence on our perception (Kühnen et al., 2001; Kühnen and Oyserman, 2002; Lin et al., 2008; Oyserman and Lee, 2008; Oyserman et al., 2009). However, to date, research on ASD and priming is sparse. Thus, the aim of the current study was to test, whether a processing preference can be overcome through priming in ASD as in TD.

Some studies have already employed priming tasks in autistic people. Semantic priming makes use of a semantic relationship between the content being primed and the task that follows. It has been used in some studies to explore the spontaneous use of a contextual, i.e., global, processing mechanism in autistic and TD individuals (Kamio and Toichi, 2000; Toichi and Kamio, 2001; López and Leekam, 2003; Hala et al., 2007; Kamio et al., 2007). Toichi and Kamio (2001) examined the ability of autistic individuals to use semantic connections to complete words. The semantic priming was performed by presenting a semantically related word to the target word. Autistic and TD individuals showed comparable abilities in completing target words. As in the TD group, the autistic individuals performed worse if there was no semantic connection between the priming word and the target word. Thus, all participants, irrespective of diagnosis, showed no differences in receptivity to a semantic priming (Toichi and Kamio, 2001). Interestingly, Kamio and Toichi (2000) differentiated in their study between two different forms of semantic priming: they used either related words or images as references to the target word to be completed. While control persons showed a comparable improvement after both priming methods, autistic individuals were significantly better when image priming was used compared to word priming. Kamio et al. (2007) examined autistic individuals without speech development delay in a similar paradigm. Using semantically close words, a priming was performed, in which participants were asked to decide whether a target word is an existing word or not. While control persons performed better after priming, no priming effect was observed in the autistic participants. This contrasts with the previously mentioned studies of the same team, rendering the literature on semantic priming in autism inconclusive.

While semantic priming lends itself to language studies, the present study set out to investigate procedural priming, which assumes that the use of a cognitive procedure on a given task can heighten the likelihood that the same procedure being used on a subsequent task. To test the modifiability of a default processing style we made use of a figure-disembedding task, in which autistic individuals have previously been shown to have a superior performance to TD individuals (Falter et al., 2008). This task requires participants to detect a smaller shape as a constituent part of a more complex figure. For procedural priming preceding the figure-disembedding task, we used the Navon task (Navon, 1977) requiring either local or global processing, as already used for priming purposes before (Förster et al., 2008). We therefore investigated whether procedural priming of global processing (henceforth: global priming) might deteriorate performance on a figure-disembedding task in autistic individuals compared to their baseline performance, which has been found to show a local processing tendency in previous perceptual studies (e.g., Falter and Bailey, 2011; Falter, 2012; Falter et al., 2013). We also investigated whether procedural priming of local processing (henceforth: local priming) might improve performance in TD individuals compared to their baseline performance. The potential of dynamically changing processing styles on figure-disembedding would show the extent to which the alleged processing differences in ASD compared to TD are modifiable by context and task design.

First, in concordance with findings by Muth et al. (2014), we hypothesized (H1) that the performance on the figure-disembedding task would be superior in autistic compared to TD individuals in the unprimed condition (i.e., baseline). Second, given the idea of a general default tendency toward a local processing style in ASD and global processing style in TD, we predicted (H2) reduced performance after global priming in the autistic group and (H3) increased performance after local priming in the TD group, whereas priming modes that already fit with the general tendency should not make a difference in performance in either group.

Descriptive statistics can be found in Table 2.

The support for H1, namely superior accuracy in participants with ASD in the unprimed baseline condition corroborates previous findings (e.g., Shah and Frith, 1983; De Jonge et al., 2006; Pellicano et al., 2006; Brosnan et al., 2012; Muth et al., 2014). However, other studies also show equal performance (e.g., Rodgers, 2000; Bölte et al., 2007; Kaland et al., 2007; Chen et al., 2008; White and Saldaña, 2011) as summarized in a recent systematic review (Horlin et al., 2016). Considering this discrepancy in findings, the results of the current study that control participants’ performance can be improved by training is highly relevant and any findings of superior or equal performance in figure-disembedding needs to be evaluated on the basis of potential differential impact of training on group averages.

Against our expectations (H2), we did not find a deterioration of performance after global or local priming in both groups (H3). A performance improvement trend in the TD group irrespective of priming points to a training effect, which was indeed confirmed in a multilevel model analysis. This was not visible in the RT data, possibly due to the fact that our instruction to participants was “as quickly but as accurately as possible” is highlighting the importance of accuracy.” Notably, neither a performance improvement nor decrease was observable in the ASD group, which renders the ASD group immune against either priming or training effects in this task. It is possible that either ASD performance is very stable and robust or that ASD individuals perform at ceiling und we therefore cannot observe a training effect. In the case of the first possibility, this would be in concert with the literature showing that ASD individuals tend to take context less into account (i.e., WCC or EPF theories mentioned in the introduction).

Our results are in accordance with previous studies that did not find any priming effect in ASD employing semantic priming (Kamio et al., 2007) but contrast others that did show semantic priming effects (Kamio and Toichi, 2000). In addition, the type of priming used in the current study, has been used in previous research (Förster et al., 2008) and was effectively a sequential priming task, for which it has been shown that target-level information (i.e., global or local priming in hierarchical stimuli) mediate priming effects in subsequent trials (Kim et al., 1999). Thus, not just the priming type but also the priming procedure used in the current study was markedly different from the semantic priming tasks described above. Yet, our results are also in line with previous experiments showing a lack of configural priming in individuals with ASD (Behrmann et al., 2006) with a sequential and non-semantic approach.

We found a significant influence of IQ on performance and improvement by training in the TD group only. Arguably, the higher the cognitive abilities the faster improvements by learning occur. Interestingly though performance in the ASD group was stable throughout. A further significant influence was found by age with decreasing performance the older the participants were. The effect of age could not have influence our findings of superior baseline figure-disembedding in ASD, because our ASD group was somewhat older than the TD control group. Thus, our results could if anything be regarded as conservative and potentially decreasing the group difference.

The findings of this study have to be interpreted in light of some limitations. First, the number of trials per condition in the figure-disembedding task was limited. However, increasing the trial number would potentially cause fatigue in participants. Future research should attempt to replicate the findings with larger trial numbers and possibly a between-subject design regarding the priming factor. Second, we provided feedback on the disembedding task, which might have contributed to the training effect. However, the feedback was unique to each trial and not strategy based, therefore we do not believe that knowing whether or not one was correct or incorrect, would have affected their performance. At most, it could have (de)motivated participants making errors over time. Third, we did not perform a sample size estimation, due to the fact that there were no previous effects to base this on and it is possible that we had too little power to detect priming effects in this study. Future research should attempt to replicate the procedure with larger samples. Further, the study sample were adults with ASD without intellectual disabilities limiting generalizability within the spectrum given that up to 50% of autistic disorders have an intellectual disability (Charman et al., 2011). Finally, the use of NF for procedural priming can be implemented in different ways. In the current study, we used a previously used approach (Förster et al., 2008). In other versions of priming with NF participants had to search at the local or global level to respond. For example, Tan and colleagues (Tan et al., 2017) did not use target letters, but merely instructed participants to attend to either the large or the small letters (see also Macrae and Lewis, 2002; Förster and Higgins, 2005). In yet another version (Gao et al., 2011), two Navon stimuli were presented to participants simultaneously in the left and right visual field, respectively, and participants were instructed to match the two displays, focusing either on the large letters, ignoring the small letters (global level), or on the small letters, ignoring the large letters (local level). Future research should attempt to replicate our findings using other procedural priming tasks, in particular Navon variations.

In conclusion, figure-disembedding seems to be a function of visuo-spatial cognition that shows solid high performance in individuals with ASD without modifiability through priming or training. In contrast, performance in control participants was improved to the level observed in ASD by training. These results imply that any findings of superior performance in figure-disembedding tasks in ASD should take the potentially differential influence of training into account and carefully examine whether the initially found superiority effect holds throughout the experimental trials performed or might lead to equal performance between groups.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethics Committee of the Medical Faculty of the University of Cologne. The patients/participants provided their written informed consent to participate in this study.

CF and AG conceptualization; LH data curation; CB, CF, AG, LH and MR formal analysis; KV funding acquisition; CF, AG, LH investigation; CF and AG methodology; CF, AG, and KV project administration; KV resources; CF, AG, and KV supervision; CF, LH, and MR original draft; CF, AG, MR, CB and KV review & editing. All authors contributed to the article and approved the submitted version.

CB, MR, and CF were supported by the DFG (German Research Council; grant numbers: FA 876/3–1 and 876/5–1). KV was supported by the DFG (CRC 1252) and by the EC, Horizon 2020 Framework Programme, FET Proactive (Project VIRTUALTIMES; Grant agreement ID: 824128).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.