Organizational benefits of neurodiversity: Preliminary findings on autism and the bystander effect

Although the bystander effect is one of the most important findings in the psychological literature, researchers have not explored whether autistic individuals are prone to the bystander effect. The present research examines whether autistic employees are more likely to report issues or concerns in an organization's systems and practices that are inefficient or dysfunctional. By bringing attention to these issues, autistic employees may foster opportunities to improve organizational performance, leading to the development of a more adaptive, high performing, and ethical culture. Thirty‐three autistic employees and 34 nonautistic employees completed an online survey to determine whether employees on the autism spectrum (1) are more likely to report they would voice concerns about organizational dysfunctions, (2) are less likely to report they were influenced by the number of other witnesses to the dysfunction, (3) if they do not voice concerns, are more likely to acknowledge the influence of other people on the decision, (4) are less likely to formulate “elaborate rationales” for their decisions to intervene or not, and (5) whether any differences between autistic and nonautistic employees with regards to the first two hypotheses, intervention likelihood and degree of influence, are moderated by individual differences in camouflaging. Results indicate that autistic employees may be less susceptible to the bystander effect than nonautistic employees. As a result, autistic employees may contribute to improvements in organizational performance because they are more likely to identify and report inefficient processes and dysfunctional practices when they witness them. These preliminary findings suggesting potential benefits of neurodiversity in the workplace are promising. However, further research is required.


INTRODUCTION
Autism is characterized by difficulties in social interaction and communication, as well as restricted and repetitive behaviors (American Psychiatric Association, 2013).Heightened attention to detail and to sensory experiences are also commonly exhibited (Lord et al., 2018).The number of adults diagnosed with autism has increased dramatically in the past 15 years.The global prevalence varies greatly but is $1% (Elsabbagh et al., 2012;Zeidan et al., 2022) with 1.8% in men and 0.2% in women.
Despite the challenges of autism, there is increasing interest amongst both public and private sector employers to harness the potential benefits of this largely untapped talent pool.Autism is a neurodevelopmental condition (Lord et al., 2018); more specifically, compared with nonautistic controls, autistic participants have shown atypical neural activation in functional MRI studies when processing language and faces (Solomon-Harris et al., 2022), observing naturalistic social situations (Pantelis et al., 2015), perceiving biological motion (Kana et al., 2009), mentalizing about someone else (Lombardo et al., 2011), or making moral decisions (Hu et al., 2021).Because of this diversity and variance in brain function and cognition, autism is considered to be a neurodivergent condition (Hoogman et al., 2022).As a result of this neurodiversity, autistic employees may bring new perspectives to a company's efforts to create or recognize value.
Potential organizational benefits of neurodiversity include reputational enhancement, productivity gains, quality improvements, boosts in innovation capabilities, and broad increases in employee engagement (Austin & Pisano, 2017).There are some qualitative reports in the literature suggesting advantages of neurodiversity in the workplace.For example, Cope and Remington (2021) asked autistic adult employees their opinions about employment-related strengths that they experienced.Laboratory experiments exploring attention to detail in a visual search task and tolerance for repetitive tasks have also been reported (e.g., Gonzalez et al., 2013).Overall, however, the breadth and quality of evidence reported in the literature thus far is insufficient to provide clear support for a workplace autism advantage (see Bury et al., 2020 for a review) This study considers one important way in which autistic employees may contribute to improvements in organizational performance because of an increased willingness to identify and report inefficient processes and dysfunctional practices.

Autism and the bystander effect
According to the bystander effect, when witnessing situations that are inappropriate or harmful, the likelihood of intervening and its promptness decreases with increasing group size (Darley & Latane, 1968).This principle of social inhibition is one of the most widely replicated research findings in the psychological literature (see Fischer et al., 2011 for a review).Three main mechanisms have been identified to explain this phenomenon (Ross & Nisbett, 2011): (1) diffusion of personal responsibility (believing someone else will intervene), (2) group influence (relying on others' reactions to recognize the need for help; perception that other witnesses do not appear to be worried), and (3) evaluation apprehension (prevailing social norms about exercising "voice"; lack of confidence in one's competence to intervene; fear of being negatively judged by others if they misinterpret the situation and there is no need to help).We propose that autistic employees are less likely to be influenced by these causal factors and, as a result, are more likely to identify and report dysfunctional organizational practices or inefficient organizational processes when they witness them.
Our work contributes to the literature on neurodiversity in several important ways.First, there is no research on whether autistic individuals are prone to the bystander effect.Prior literature has investigated differences in susceptibility to social influence in conformity between autistic and nonautistic individuals, but these studies have yielded conflicting results.One study with children (Yafai et al., 2014) found that autistic children conformed less frequently than nonautistic children.A similar association of autistic traits and lower sensitivity to "antisocial" peer influence (when peers were selfish and did not cooperate) was also reported in a study assessing adolescents (Van Hoorn et al., 2017).However, two studies involving adults concluded that autistic and nonautistic adults are equally susceptible to social influence (Bowler & Worley, 1994;Lazzaro et al., 2019).Lazzaro et al. (2019) suggest that the different results between studies with children and adolescents and studies with adults may be due to the different ages of autistic participants.In other words, the extent to which autistic individuals exhibit a tendency to conform may change with age.Autistic adults may acquire social conformity as a social strategy (i.e., "camouflaging" a construct which we will consider below), whereas autistic children have yet to develop such a strategy.
In addition, some studies have shown that autistic people are less susceptible to social evaluation by others when making moral judgments (Frith & Frith, 2011).For example, when asked to make charitable donations in the presence or absence of others, nonautistic participants donated significantly more in the observer's presence than absence.In contrast, autistic participants were not affected by the presence of an observer in this task (Izuma et al., 2011).Similarly, evidence indicates that autistic children and adults are less responsive to notions of reputation management compared with their neurotypical counterparts (Cage et al., 2013).

Causal reasoning and false beliefs
Furthermore, we propose that even if autistic individuals are susceptible to social influence, they will be less likely than nonautistic individuals to formulate elaborate "rationales" for their decision not to intervene.In other words, we hypothesize that autistic employees will be less likely to espouse false beliefs about whether they were or were not influenced by the presence of other people when deciding to report inefficient processes or dysfunctional practices that they witness in an organizational setting.
We propose this hypothesis because of evidence indicating that when people make judgments about how a particular stimulus (in this case the presence of other people) influences a particular response (in this case helping behavior), their judgments are typically based on a priori causal theories ("causal schemata," psychological rules, or cultural beliefs) describing likely stimulus response relations (Nisbett & Wilson, 1977).People on the spectrum may have failed to learn these "causal schemata" as infants during the critical developmental window (Baron-Cohen et al., 1985).This is the requisite period of neural plasticity for many relevant developmental experiences including joint attention which is associated with differences in language development (both expressive and receptive) in autism (see Bottema-Beutel, 2016 for a review).The social and communication challenges experienced by people on the spectrum are attributed in part to lack of understanding of attention in others as infants (Bruinsma et al., 2004).Understanding of self and others' intentions is a critical precursor of "theory of mind," the ability to predict and explain an agent's behavior in novel circumstances based on an understanding that agents behave, not with respect to reality, but with respect to their beliefs about reality.We propose that autistic individuals may not have learned these culturally supplied beliefs or explanations about the causes of behavior and, as a result, they may be less susceptible to the self-serving cognitive distortions and rationalizations that accompany most people's attempts to report on their actions and decisions.
The influence of false beliefs is clearly seen in studies on the effects of the presence of others on helping behavior.Subjects in the bystander effect experiments conducted by Latane and Darley (1970) were routinely asked during the debrief whether they thought they had been influenced by the presence of other people.Subjects persistently claimed that their behavior was not influenced by the other people present, irrespective of whether the researchers asked the question subtly, directly, tactfully, or bluntly.Even when the researchers explained that the presence of others does inhibit helping and this result has been replicated in many experiments in a wide variety of settings, this denial persisted (e.g., "Maybe that is true for most other people, but not for me.").
Other researchers working in the dissonance tradition (e.g., insufficient justification and attribution) also asked their subjects about their thought processes, with the same result.As in the case of awareness of the influence of other people on helping behavior, subjects seem unable to accurately report about the effects of stimuli on complex, inferential responses (e.g., Aronson & Mills, 1959;Freedman, 1963;Nisbett & Schachter, 1966;Zajonc, 1968).Typically, people believe that while others may fall prey to such influences (e.g., the influence of the number of witnesses on the decision to intervene), they somehow are immune to these influences.As suggested by Nisbett and Wilson (1977) in their classic review paper, many different experiments have been conducted in a wide variety of settings (cognitive dissonance and attribution research, induced compliance, etc.) where nonautistic subjects seemed to be utterly unaware of the influence of situational and other factors on their behavior.Drawing on these observations, we posit that autistic people are less likely to subscribe to elaborate rationales or "false beliefs" because the implicit theories about the causes of human behavior that these elaborate rationales are based on were not learned during the formative years of development.
Studies of "moral dumbfounding" (i.e., where people remain committed to moral judgments despite an inability to provide justifications to support them) provide further evidence suggesting differences between autistic and nonautistic individuals.This research has shown consistent differences in moral judgments made by autistic adults without intellectual disabilities compared with nonautistic adults when considering complicated moral decisions (Bellisi et al., 2018;Buon et al., 2013;Fadda et al., 2016;Hu et al., 2021;Margoni & Surian, 2016).Specifically, autistic people are more inclined to adopt a utilitarian approach in which moral actions and rules focus on maximizing wellbeing.Nonautistic adults, by comparison, tend to employ a deontological approach in which moral actions are those which, irrespective of consequence, comply with certain categorical rules. 1 As convincing as these moral rules appear to be, however, they may not accurately describe the reasons underlying the choice.For example, nonautistic people tend to base their judgments of moral acceptability of behavior on their emotional response to that behavior whereas autistic people rely more on simple rules and consideration of potential adverse consequences of the behavior when judging moral acceptability (Brewer, Happé, et al., 2015;Brewer, Marsh, et al., 2015).In other studies, autistic people have been found to make similar moral judgments as nonautistic people, but autistic participants generated fewer sophisticated or elaborate rationales based on abstract principles when explaining why the transgression was wrong (Bellisi et al., 2018;Shulman et al., 2012).
This dynamic is clearly evidenced in studies of sacrificial dilemmas, the most famous of which is the "runaway trolley" (Gleichgerrcht et al., 2013).On average, 80% of nonautistic participants will push a switch to divert a runaway train, sacrificing one life in order to save five lives 1 One of the authors (Braxton L. Hartman), who is autistic, points out that utilitarianism allows for exceptions and extenuating circumstances, whereas deontology does not.This seems to cut against the prevailing stereotype of autistics being fixated on rules and unable to deviate.
(the impersonal condition).However, only 10%-15% will push a person off a footbridge to stop the train (the personal condition), again sacrificing one life in order to save five lives.When nonautistic people are asked to explain their different decisions in this dilemma, they almost always misinterpret if not hide the core premises and processes that led to their conclusion (Hauser et al., 2007).In fact, most participants fail to provide any plausible justification for the different decisions they make in the personal and impersonal condition.Instead, they draw on general knowledge and abstract moral conceptions (beliefs about right vs. wrong) in the effort to explain what was an intuitive, emotional, and unconscious response.
This dual process theory of motivated reasoning (e.g., Haidt, 2001;Kahneman, 2013) does not appear to have the same degree of influence on moral decisions made by autistic people.Thus, in the runaway trolley problem, 80% of autistic participants are willing to push the switch in the impersonal condition, just like nonautistic participants.However, 40% of autistic participants will push the person off the footbridge in the personal condition, considerably more than the 10%-15% found for nonautistic participants (Gleichgerrcht et al., 2013).

Theory of mind
The utilitarian approach adopted by people on the spectrum when making complex decisions, as seen in the trolley problem, has been attributed by some to deficits in empathy (Gleichgerrcht et al., 2013).Empathy, the ability to recognize and understand the state of mind of others including their beliefs, desires, and particularly others' emotions, is a concept that is related to theory of mind, the ability to attribute mental states to oneself and others.It is important to note however that deficits in empathy are not a central feature of autism; lack of empathy is neither necessary nor sufficient for an autism diagnosis.In other words, there are autistic individuals without deficits in empathy and individuals with deficits in empathy who are not autistic.Evidence suggests that autism is associated with atypical theory of mind but not with deficits in empathy (Brewer, Happé, et al., 2015;Brewer, Marsh, et al., 2015;Butera et al., 2023).In fact, autistic children show greater awareness of their own emotions than nonautistic children over a long period of time (Murphy et al., 2017).
Atypical theory of mind may in fact be an advantage in ethical reasoning.Utilitarian ethical philosophy is characterized by impartiality and agent-neutrality, that is, everyone's happiness counts the same.When one maximizes the good it is impartially considered.According to this perspective, the so-called empathy deficit may in fact be preventing bias from impeding impartiality and would therefore be a strength rather than a deficit.In addition, studies have found that, compared with nonautistic children, autistic children are less likely to engage in deception and sabotage (Sodian & Frith, 1992), less likely to manipulate their behavior and recognize when others are being manipulative (Yirmiya et al., 1996), less likely to spontaneously tell lies in order to conceal information (Talwar et al., 2012), and are more likely to adhere to rules (Vicente & Falkum, 2023).These differences have been attributed to atypical theory of mind (Baron-Cohen, 1992) and lend additional empirical support to the reconceptualization of this deficit as an ethical strength.
Theory of mind is acquired sequentially.First, we learn to recognize that others have different desires, then different beliefs, then different knowledge and then finally that others have false beliefs and capabilities of hiding their emotions (Wellman & Liu, 2004).Children come to an understanding of false beliefs as a result of their continuing experiences coordinating mental states with others, especially in the context of social interactions and communication that requires them to compare their respective perspectives (Tomasello, 2018).Different people may develop more or less effective theory of mind (Frith & Happé, 1994).
It is not until age 4 or 5 that children show much understanding of beliefs (see Wellman et al., 2001 for a review and meta-analysis).At this age, children first recognize that people may be mistaken in their beliefs about the world, as evidenced by their successful performance on experimental false belief tests.In the classic Sally-Ann task (Wimmer & Perner, 1983), an example of a false belief test, the child must predict where someone will look for an object that the child sees moved to a new location, but the person does not see it moved.Most children pass this false belief test by age 4 (including, e.g., those with Down's syndrome) but 80% of autistic children are unable to do so (Senju et al., 2009), even though they can successfully answer the control questions ("Where is the object now?" and "Where was it placed originally?").While numerous studies have demonstrated atypical theory of mind in autistic individuals as observers (see Senju, 2012 for a review), the experimental tasks employed in this research do not place participants in actual social interactions.Some investigators are now beginning to study the dynamics of social interactions between autistic and nonautistic people.In this context, theory of mind is viewed as a two-way problem between differently disposed actors.This has been described as the "double empathy problem" (Milton, 2012).

Camouflaging and susceptibility to social influence
Finally, we propose that the potential advantages of neurodiversity may be moderated when autistic people are attempting to minimize the visibility of their autism during social interactions.Social camouflaging describes an explicit effort to "mask" or compensate for autistic characteristics during social interactions by modifying one's behavior in social situations in order to present a less visibly autistic persona (Hull et al., 2017).It is motivated by the desire to reduce discrimination, smooth social interactions, and achieve success in employment or education.A three-factor model of camouflaging has been identified: (1) compensation (strategies used to compensate for social and communication difficulties), for example, "use social skills learned from watching others' interactions," (2) masking (strategies used to present a nonautistic or less autistic persona to others), for example, "adjust face and body to appear relaxed," and (3) assimilation (strategies used to fit into uncomfortable situations), for example, "feel the need to put on an act" (Hull et al., 2019).Camouflaging requires a structured approach to learning and modeling which may involve observing (mimicking) nonautistic peers.We speculate that autistic participants may be more susceptible to social influence if they are camouflaging and, as a result, be more prone to the bystander effect.

Hypotheses
We predict that autistic employees (1) are more likely to report they would voice concerns about organizational dysfunctions, (2) are less likely to report they were influenced by the number of other witnesses to the dysfunction, (3) if they do not voice concerns, they are more likely to acknowledge the influence of other people on the decision, (4) they are less likely to formulate "elaborate rationales" for their decisions, and (5) that any differences between autistic and nonautistic employees with regards to the first two hypotheses, intervention likelihood and the influence of others, will be moderated by individual differences in camouflaging.

Participants
Thirty-five employed adults with a clinical diagnosis of autism were recruited through social media as well as direct contact with autism organizations and charities.Autistic participants self-reported an official autism diagnosis from a qualified healthcare professional and were asked to detail the label of diagnosis (e.g., Autism, Asperger's Syndrome, Pervasive Developmental Disorder), the age they were diagnosed, and the type of healthcare professional who diagnosed them.Diagnostic identification was further confirmed using a 10-item version of the Autism Spectrum Quotient (AQ) for adults (Baron-Cohen et al., 2001;Lundqvist & Lindner, 2017;Stewart et al., 2015).Two participants with low AQ scores (<2.0) were removed, resulting in 33 autistic participants in the data analyzed below (10 males and 23 females, mean age = 36.12years, average length of employment = 1-5 years, mean AQ score = 3.9).Average age of diagnosis for autistic participants was 30.1 years; this variable follows a normal distribution with a mean past the mode due to a very high maximum value.Most diagnoses were provided either by clinical psychologists (17/33) or psychiatrists (9/33).
Fifty-one nonautistic employed adults were recruited using social media and university students (in a business school and medical faculty in two large North American universities).Seventeen participants with high AQ scores (2.8 and above) were removed and one participant with missing data was removed, resulting in 34 nonautistic participants (18 males, 16 females, mean age = 22.53 years, average length of employment = 1-5 years, mean AQ score = 2.2).For both groups, specific data on ethno-racial identity, socioeconomic status, and educational attainment levels were not recorded (more detailed description of participants' demographic characteristics are available in the Figures B1-B6).

Procedure
Participants were given an online link to the study, hosted by Qualtrics, where they read the information sheet and provided informed consent.In addition to demographic questions and the AQ, participants also completed the Camouflaging Autistic Traits Questionnaire (Hull et al., 2019) and the Moral Disengagement Survey (Detert et al., 2008).Scores for Camouflaging and Moral Disengagement were entered in the regression analyses conducted to determine the ideal model for ANOVAs (analyses of variance) testing differences between autistic and nonautistic participants on the main dependent variables.
Finally, the four dependent variables (intervention likelihood, degree of influence, false beliefs about the influence of others, and rationale for decisions) were assessed using the Organizational Scenarios Survey which was developed to assess how people reason about workplace situations when they witness practices or actions that may be problematic and have negative consequences in the form of inefficiencies, inequities, quality defects or ethical concerns.Participants read seven short scenarios describing workplace vignettes, each of which contained an example of an organizational dysfunction, either an ethical issue or an operational inefficiency.Each scenario also contained a specific number of other bystanders or individuals who are present ranging from one other person to 12 other people.These scenarios were initially developed by conducting critical incident technique interviews (Flanagan, 1954) with "subject matter experts" (SMEs).These SMEs were employed autistic adults or managers of employed autistic adults. 2  Following each scenario, participants were asked to indicate how likely it is that they would take action to address the issue in the scenario (Intervention Likelihood) on a scale from 1 to 4, where 1 represents "not at all likely" and 4 represents "very likely."Participants were also asked to explain why they decided to intervene or not in each scenario (Rationale for Decisions).Rationales were classified into one of two categories, "concrete" or "abstract."Rationales were categorized as concrete when participants cited a simple rule that prohibits the behavior in question (e.g., states that it is unethical) or considered the results, outcomes, or consequences of acting/not acting in terms of external stakeholders like customers or regulators.Rationales were classified as abstract when they considered the circumstances (e.g., whether it is their responsibility) or the potential impact on internal stakeholders (e.g., potential to embarrass or upset management).All explanations were double-scored by two raters who were blind to group membership.Based on interrater agreement calculations using Cohen's kappa (Landis & Koch, 1977), three of the seven scenarios were not included in the analysis because of low kappa coefficients (<0.20).The final version of the survey consisted of the remaining four scenarios (all seven scenarios, including the ones that were not used, are provided in Appendix A).The final kappa coefficient was 0.379 (0.416 for autistic participants and 0.233 for the nonautistic participants).When the two raters coded a rationale differently, then a third rater, also blind to group membership, made a consensus pick (either "concrete" or "abstract").The number of other witnesses or people present in the final four scenarios used in the analyses ranged from 1 to 10.After explaining the rationale for their decisions to act or not, participants were asked to what degree, if any, their decision was influenced by the presence of other people (Degree of Influence) where 1 represents "not at all" and 4 represents "to a great extent."On average, the battery took 20 min to complete.

Statistical analysis
For Hypotheses 1 and 2, we first conducted step-wise backward regression of mixed-effects models which included potential confounding factors (i.e., age, sex, length of employment, hours worked per week, organization type, AQ, moral disengagement, and camouflaging) in order to create new, simpler versions of the model with fewer random effects (potential confounding factors), and compare the models to determine whether including each effect actually improves the model, that is, explains more of the variance.The best model was the one with the lowest Akaike Information Criterion (Vrieze, 2012).When the final model had significant fixed effects, we then tested for differences between autistic and nonautistic participants on the first two dependent variables (intervention likelihood and degree of influence) with ANOVAs (the regression model results and ANOVA summary tables for Hypotheses 1 and 2 are displayed in the Tables B1-B4).Degrees of freedom on the optimized model vary because the optimized models differ in how many variables they ended up with.For testing Hypothesis 3, we conducted an ANCOVA with autism diagnosis as the covariate followed by correlations (for the ANCOVA summary table see Table B5).Chi-square tests were used for Hypothesis 4 to analyze rationale for decisions, a categorical variable.For Hypothesis 5, we conducted an ANCOVA to test the interaction between camouflaging and the effects reported for Hypotheses 1 and 2 (intervention likelihood and degree of influence), followed by correlations between camouflaging and each of the effects, and then finally a t-test to determine if the correlations are significantly different for autistic and nonautistic participants (for these ANCOVA summary tables see Tables B6 and B7).Actual p-values are reported.The p-values < 0.05 are considered to be statistically significant.All analyses were done using R Core Team (2013).

RESULTS Hypothesis 1: Intervention likelihood
We proposed in Hypothesis 1 that autistic employees will be more likely than nonautistic employees to report they would voice concerns when they witness instances of organizational dysfunction or inefficient processes.Figure 1 shows the means and standard deviations for participants' ratings of "how likely are you to intervene?" on a 4-point Likert scale after being presented with each of the four scenarios.
After pooling responses across the four scenarios, the likelihood of intervention differed significantly between autistic and nonautistic groups, F (1, 245) = 16.056,p = 0.00008.Thus, Hypothesis 1 is supported; specifically, autistic employees are more likely than nonautistic employees to report they would voice concerns when they see something wrong happening in an organization.

Hypothesis 2: Degree of influence
In Hypothesis 2, we expected that autistic employees would be less likely to report that they were influenced by the number of other witnesses to the dysfunction than nonautistic employees.Figure 2 shows the means and standard deviations for participants' ratings of the degree to which they were influenced by the presence of others when estimating their likelihood of intervening.The same statistical procedure was employed as in Hypothesis 1: step-wise regression to identify the best fit model followed by an ANOVA to test the difference between autistic and nonautistic groups.After pooling responses across the four scenarios, there was a significant difference between autistic and nonautistic participants in the degree to which they reported they were influenced by the presence of others, F (1, 54) = 5.366, p = 0.0244.Thus, Hypothesis 2 is also supported; specifically, autistic employees are less likely than nonautistic employees to report they were influenced by the presence of others when deciding whether to voice concerns when they see something wrong happening in an organization.

Hypothesis 3: Acknowledging the influence of others
The third hypothesis predicts that autistic employees are less likely to espouse false beliefs about whether they were or were not influenced by the presence of others.Specifically, if autistic employees do not voice concerns, they should be more likely to acknowledge the influence of other people on the decision.In other words, as the likelihood of intervening decreases, autistic participants should be more likely than nonautistic participants to acknowledge that they were influenced by the presence of others when deciding not to intervene.Thus, we predict a significant negative correlation between intervention likelihood and influence of others for autistic participants and no correlation or at least a less significant correlation for nonautistic participants.
To determine if autistic and nonautistic participants have different correlation slopes between likelihood of intervention and influence of others, we did an ANCOVA with autism diagnosis as the covariate.The interaction was significant, F (3, 264) = 12.03, p = 0.00000002.Accordingly, we then conducted Spearman's correlations for each group (the correlation slopes are shown in Figure 3).The correlation between likelihood of intervening and degree of influence is significant for both autistic participants (r s = À0.33,p = 0.0001) and for nonautistic participants (r s = À0.17,p = 0.05).However, the difference between these two correlations is also significant (t = À2.126,p = 0.03) and in the predicted direction, that is, compared with nonautistic employees, autistic employees are more likely to acknowledge the influence of others on their decisions to intervene as their likelihood of intervening decreases.F I G U R E 3 Correlation between likelihood of intervention and degree of influence.

Hypothesis 4: Rationale for decisions
In Hypothesis 4, we predicted that autistic employees are less likely to formulate "elaborate rationales" (i.e., abstract explanations vs. concrete explanations) for their decisions than nonautistic employees.Since rationale is a categorical variable, we test whether autism influences rationale through a simple Chi-square test of Independence.Results are shown in Figure 4.The frequency of abstract to concrete rationales amongst autistic and nonautistic employees differed only for the scenario in which there were 10 bystanders (χ2 = 9.8465, df = 1, p = 0.002), with autistic participants showing a higher concrete-to-abstract ratio (26 concrete vs. 7 abstract) than nonautistic participants (14 concrete vs. 20 abstract).For the remaining three scenarios, there is not a difference between autistic and nonautistic employees in the ratio of concrete-to-abstract explanations.

Hypothesis 5: Moderating influence of camouflaging
As expected, camouflaging scores were significantly higher amongst autistic (mean = 3.7) than nonautistic participants (mean = 3.0), F (1,64) = 18.901, p = 0.00005.However, this hypothesis focuses on whether the differences found between autistic and nonautistic employees with respect to the first two hypotheses (intervention likelihood and the influence of others) change as a result of the level of camouflaging reported.
Contrary to expectation, camouflaging increases intervention likelihood.However, the influence of camouflaging on intervention likelihood is not significantly different for autistic and nonautistic employees, F (1, 261) = 0.783, p = 0.377.Similarly, camouflaging does not have a differential effect on degree of influence ratings for autistic and nonautistic employees, F (1, 261) = 0.9385, p = 0.334.However, higher levels of camouflaging increase the degree to which autistic employees report they are influenced by others when deciding to intervene (r s = 0.211, p = 0.015) whereas this same effect is not found for nonautistic employees (r s = 0.12, p = 0.891).

DISCUSSION
The purpose of this study was to determine whether autistic employees are more likely to identify and report dysfunctional practices and inefficient processes that they witness, thereby contributing to potential enhancements in the organization's competitive advantage.Specifically, we provide evidence that compared with nonautistic employees, autistic employees are (1) more likely to indicate they would identify and report inefficient processes and dysfunctional practices in the organization when they witness them; (2) less likely to report they were influenced by the presence of other bystanders when deciding whether they would intervene when they witness instances of inefficient processes or dysfunctional practices; (3) more likely to acknowledge the influence of others when they decide not to intervene; (4) less likely to formulate elaborate, abstract explanations for their decision to intervene when there are a large number of bystanders present, and (5) more likely to acknowledge the influence of others on their decisions to intervene when they report higher levels of camouflaging.The results of this preliminary investigation on the organizational benefits of neurodiversity highlight key theoretical and organizational implications.

Theoretical implications
Our results extend the current theory in several ways.This is the only study we are aware of that has F I G U R E 4 Ratio of concreteto-abstract rationales for each of four scenarios.
specifically considered whether autistic individuals are less prone to the bystander effect than nonautistic individuals.This research provides new insights into understanding how autistic people process information and makes decisions differently than nonautistic people.Based on these preliminary findings, autistic people appear to be less susceptible to self-serving cognitive distortions and rationalizations that accompany most people's attempts to report on their actions and decisions.This may be because as young children they did not learn the culturally supplied explanations for behavior normally acquired during this critical period of neuroplasticity early in development (Buon et al., 2013).As a result, in this study, autistic employees were less likely to report that they were influenced by the presence of others when deciding to intervene and, when they reported that they were not likely to intervene in a situation, they were less likely to espouse false beliefs about whether they were or were not influenced by the presence of others when deciding to intervene.
Unexpectedly, we found that likelihood of intervention increases as camouflaging increases.This somewhat surprising result may reflect the focus of this research on functioning in the workplace.In the workplace context, the reasons for camouflaging may have more to do with performing well at one's job, for example, demonstrating that you are a responsible person, than easing everyday social interactions and relationships.If this is the case, then higher levels of camouflaging might raise the likelihood of intervening.This interpretation is consistent with findings indicating two main reasons for camouflaging: (1) relational reasons to ease everyday social interactions, and (2) conventional reasons serving a functional purpose such as in workplace or educational contexts (Cage & Troxell-Whitman, 2019).In addition, Cage and Troxell-Whitman (2019) also found that autistic women and those diagnosed in adulthood were more likely to endorse conventional reasons for camouflaging relative to relational reasons in comparison with autistic men and those diagnosed in childhood.The demographic characteristics of our sample of autistic participants (on average diagnosed at 31 years of age and 70% female) are consistent with this finding.
Although camouflaging did not differentially influence the decision to intervene between autistic and nonautistic participants, we found that higher camouflaging resulted in increased recognition of the influence of others for autistic participants whereas the same effect was not found with nonautistic participants.These findings are consistent with results indicating a different developmental trajectory of social influence in autism (Jorgensen et al., 2020;Large et al., 2019;Lazzaro et al., 2019;Van Hoorn et al., 2017;Yafai et al., 2014).For example, Jorgensen et al. (2020) found that camouflaging tends to increase from early to late adolescence for neurotypical individuals but not for autistic participants.Also, Large et al. (2019) found that by early teens, typical neurodevelopment allows social influence to systematically bias perception processes in a visual search task.That same bias did not emerge in autistic adolescents.This extant literature suggests a distinct developmental trajectory of social influence processes in autism which we will briefly summarize: (1) Autistic children first begin to encounter challenges in social interactions and communication with their nonautistic peers in early childhood (Lord et al., 2015).( 2) In later childhood and adolescence, many autistic teenagers begin to engage in camouflaging for relational reasons, using the three camouflaging mechanisms (masking, compensating, assimilating) identified by Hull et al. (2019) to better understand and get along with their nonautistic peers.
(3) However, in adulthood, particularly for people on the spectrum who can get and maintain employment, our findings suggest that camouflaging may have more to do with performing well in their job and being seen as a responsible employee.Now, camouflaging is used to help the autistic individual recognize when others are being deceptive or are under the influence of false beliefs; in particular, the influence of self-serving cognitive distortions and rationalizations that accompany nonautistic people's attempts to report on their actions and decisions.In this context, we tentatively propose a fourth camouflaging mechanism in addition to masking, compensation, and assimilation which we call "sleuthing." We propose that sleuthing serves to detect deception and false beliefs in order to reduce susceptibility to "antisocial" influences such as the bystander effect. 3According to this model, the reduced susceptibility to the bystander effect in autism may be accounted for, in part, by higher levels of camouflaging reported by autistic adults in comparison with nonautistic adults.

Organizational implications
This increased willingness to bring attention to issues (whether the problem is a minor mistake in a training manual or gross misconduct on the part of a manager) can have important implications for organizations.From a societal perspective, there are obvious benefits to successfully employing and integrating individuals with disabilities (e.g., reducing public assistance costs, increasing tax payments).For the employer, there are public relations and marketing advantages when their organization is seen as having a commitment to equity, diversity, and 3 Albeit purely anecdotal, it is interesting to note the way in which one autistic individual, again Braxton L. Hartman, characterized their experience of these three developmental stages of social influence integration in autism: (1) In early childhood, the experience of attending school with nonautistic children was described as akin to visiting the zoo, except the animals were not in their cages.
(2) In adolescence, camouflaging strategies were described as pretending to be a secret agent who had been inserted behind enemy lines with a mission to avoid detection and blend in in a foreign environment.(3) In adulthood, the tendency for nonautistic people to deny or rationalize the influence of situational and other factors on their behavior was described as "the weird neurotypical syndrome."inclusion.Additionally, we have provided empirical findings in this preliminary study that highlight potential benefits in the form of improvements in firm competitiveness that can be garnered by employing people on the spectrum.Yet autistic adults experience rates of unemployment (and underemployment) as high as 85%-90% (Howlin, 2013;Taylor & Seltzer, 2011).These employment numbers improve significantly when autistic individuals and people with other developmental disabilities receive job search, job placement, and employment supports (Chen et al., 2015;Kaya et al., 2016;Mavranezouli et al., 2014).
Barriers to the employment of autistic adults also include hiring processes that define talent too narrowly and the use of unstructured job interviews which may be biased against people with atypical manners of interaction (Willis et al., 2021;Wilson & Bishop, 2021).Even if they get hired, negative attitudes or stereotypes about customizing standardized job roles or tailoring working conditions to accommodate special needs are other challenges to overcome (Solomon, 2020).Whether the goal is simply to support a good cause or to achieve better business results, there is a need for organizations to develop special employment practices for hiring, onboarding, and managing neurodiverse employees.
Recently, several organizations have developed more equitable and accurate ways of sourcing prospective hires, collecting information, assessing competencies, and screening candidates to identify neurodiverse people who can perform the job (e.g., Austin & Pisano, 2017).They also assist organizations with a structured onboarding process and, where necessary, provide guidance on how to modify aspects of the work environment to deal with social, sensory, and physical factors to which some autistic individuals are sensitive.Finally, educating managers and coworkers about the myths and realities of attitudes towards autistic people helps to ready the social environment for accepting neurodiversity in the workforce.Follow-up studies evaluating these approaches have reported positive findings (e.g., Khalifa et al., 2020).However, there is a need to do more research on improving nonautistic people's attitudes towards autism, especially in the workplace.
It is also important to note that although an increased likelihood of reporting organizational dysfunctions is beneficial to the organization, it may well be detrimental to the individuals themselves if they are subject to retribution (see Nicholls et al., 2021 for a review of the research literature on consequences of whistleblowing).Future work should address what organizational systems need to be implemented in order to reap the benefit of autistic employees while shielding them from potential negative repercussions.

Limitations and future research directions
Because of the relatively small sample size, we did not make corrections for multiple comparisons.Future studies should be based on larger sample sizes.In addition, we did not collect information on participants' socioeconomic status, IQ, educational level, or ethnoracial identity.Also, different recruitment methods were employed to source autistic and nonautistic participants for the study.As a result, in addition to whether participants were autistic or not, there were also differences between our two comparison groups in terms of age and sex.Our nonautistic participants are considerably younger (22.5 years of age) than our autistic participants (36 years of age) and the ratio of female to male participants is considerably higher in the autistic group (23 females, 10 males) than the nonautistic group (16 females, 18 males).Future studies exploring these effects should employ participant recruitment strategies that ensure better matching between autistic and nonautistic groups on key demographic variables.Nonetheless, there is a lack of research on the potential advantages of autism in the workplace.Therefore, the current study does add to a growing body of literature focusing on neurodiversity.
In addition, this report may be considered a pilot study for the development of a survey to measure the dependent variables of interest: intervention likelihood, influence of others, acknowledging the influence of others, and type of rationale employed.Seven different scenarios (situations or critical incidents identified by "job experts") were described and the number of other witnesses present in each scenario was fixed.However, we found that participants' ratings were also influenced by the characteristics of the scenario, not just the number of other witnesses present in the scenario. 4Based on the rationales provided by participants, three of the seven scenarios appeared to contain social dynamics and subtleties that masked or camouflaged the intended focus on whether, as a bystander in a situation where something wrong is happening, would you act?The remaining four scenarios that we used for data analysis were relatively free of any of these situational factors, as evidenced by higher kappa coefficients indicating better inter-rater agreement when coding rationales.Future research might even restrict the number of scenarios further.For example, two scenarios could be sufficient; one focused on an operational inefficiency and one focused on an ethical misdeed.Research participants would then only respond to those two scenarios but the number of witnesses present in the scenario could be varied (e.g., from 1 to 10, the range of witnesses used in this study) in a Latin square design.In addition to number of bystanders, additional empirically linked moderator variables (Fischer et al., 2011) might also be considered in the selection of scenarios (e.g., degree of danger in the scenario, high vs. low ambiguity of the situation, costs of intervention vs. nonintervention, presence vs. absence of perpetuators).
Furthermore, although we did capture participants' explanations for why they believed they would or would not take action to address the situation described in each scenario, we only considered whether those explanations were concrete or abstract.We did not consider individual differences in how participants interpreted or evaluated the scenarios, for example, how problematic, or unproblematic they found each scenario to be.Future research should explore whether individual differences in how scenarios are evaluated impacts the decision to act or not.
Finally, what was measured here were beliefs or intentions to voice concern, not actual behavior, which runs the risk of the "intention-behavior gap" (Sheeran & Webb, 2016).There is research indicating that intentions based on moral norms are better predictors of behavior than intentions aligned with attitudes (Godin et al., 2005) suggesting that the intention-behavior gap may be less of a concern in this study.In addition, while we are not aware of any research on whether the intention-behavior gap is different in autism, there is evidence indicating that, in comparison to nonautistic people, autistic people tend to prefer literal interpretations of words and intentions (Vicente & Falkum, 2023), tend to be more rule following in the social realm (Shulman et al., 2012), and are less likely to tell lies (Talwar et al., 2012).While autistic individuals may have difficulty understanding the goals and intentions behind other people's actions (e.g., Phillips et al., 1998), autistic individuals are not less likely to implement their own intentions than nonautistic individuals (see Poljak & Bekkering, 2012, for a review).Accordingly, there are reasons to believe that the autistic participants in this study may be more inclined to act out their intentions with regards to intervention likelihood than the nonautistic participants.However, the fact that we could not find any relevant research on the intentionbehavior gap in autism suggests that this is an area that needs to investigated.

F
I G U R E 1 Average ratings for likelihood of intervening for each of four scenarios.Error bars represent ±1 standard deviation.2 Braxton L. Hartman was involved as a research partner in all aspects of the study including helping to develop the Organizational Scenarios Survey as one of the SMEs.

F
I G U R E 2 Average degree of influence of others on decision to intervene for each of four scenarios.Error bars represent ±1 standard deviation.