Autistic adults show enhanced generosity to socially distant others

Abstract

Sharing resources is fundamental for human cooperation and survival. People tend to share resources more with individuals they feel close to compared to those who are more socially distant. This decline in generosity at increasing social distance is called social discounting and is influenced by both social traits and abilities, such as empathy, and non-social psychological factors, such as decision-making biases. People who receive a diagnosis of autism show differences in social interaction as well as displaying differences in non-social domains, such as more restricted and repetitive behaviours. We investigated social discounting in autism and found that autistic adults were more generous than neurotypical participants, which was driven by greater generosity to socially distant others. Crucially, we also investigated framing effects during prosocial decision-making. Autistic participants were less susceptible to whether decisions were framed as causing monetary gains, compared to preventing monetary losses, for the potential recipient. Our results support the view of ‘enhanced rationality’ in autism as participants’ prosocial decisions were less influenced by potential biasing information, such as the closeness of the recipient or how choices were framed. Therefore, the differences seen in autism, as well as posing certain challenges, can also have prosocial consequences.

Lay abstract

Autistic people show differences in their social behaviour. But how autism affects decisions to share resources, an important part of cooperation, was previously unclear. In our study, participants made decisions about how to share money with different people, including people they felt close to, such as a friend, and people they felt less close to, such as a stranger. We found that compared to a group of non-autistic participants, autistic adults shared more money overall and this was driven by greater generosity to strangers. The results suggest that autistic adults were more generous because they made fair decisions (an equal split of the money) more consistently regardless of how close they felt to the person they were sharing with. By showing that autistic adults display greater generosity, our results could help to change public perceptions of autism and potentially improve opportunities for autistic people.

Keywords

decision-making framing effect prosocial behaviour psychological theories of autism social discounting social distance

Introduction

Sharing resources is fundamental for human cooperation and survival (Nowak, 2006) and people are generally more willing to share resources with individuals they feel close to compared to those who are more socially distant (Jones & Rachlin, 2006). The extent to which generosity declines across increasing social distance is referred to as social discounting. Where greater social discounting means a greater decline in generosity towards people we feel less close to, such as strangers, compared to those we feel closer to, such as friends. Social discounting is a robust phenomenon but is influenced by a range of social traits and abilities (for a recent review, see Jones, 2022). For example, women scoring high on empathy (Olson et al., 2016) and extreme altruists (Vekaria et al., 2017) show greater generosity to strangers (i.e. reduced social discounting), and cultural differences have also been reported (Archambault et al., 2020; Ito et al., 2011; Strombach et al., 2014). Social discounting is also influenced by more general, not specifically social, psychological factors. For example, acute stress induction can enhance generosity to close others in social discounting tasks (Margittai et al., 2015), cognitive load can attenuate the decline in generosity at increasing social distance (Strombach et al., 2016) and framing generous choices as preventing a loss rather than inducing gains for another person reduces social discounting (Sellitto et al., 2021).

People who receive a diagnosis of autism show differences in social interaction and communication, such as the ability to adjust behaviour to various social contexts. In addition, autistic individuals display differences in non-social domains by showing more restricted and repetitive behaviours, often including a preference for sameness (American Psychiatric Association, 2013). Behavioural differences in autism have been linked to differences in social cognition, such as an altered sensitivity to social information (Chevallier et al., 2012), but also more general, not specifically social, aspects of cognition, such as the ability to shift attention flexibly (Pellicano, 2012). Thus, both the social and non-social aspects of autism could contribute to differences in social discounting.

Social discounting in autism

Two studies have investigated social discounting in autism and reported divergent results. Tei et al. (2019) asked autistic adults to think of people at six different social distances on a scale of 1 to 100 with the person at social distance 1 being a person they feel closest to and the person at social distance 100 being a complete stranger. On each trial, participants chose between a prosocial option which was ¥4000 for both the participant themselves and the other person, compared to a selfish option which was a greater amount for themselves (varying from ¥4000 to ¥8200 in ¥600 steps) and nothing for the other person. Tei et al. calculated the indifference point between the selfish and the prosocial option at each social distance. This is where there was a 50% chance of participants choosing the selfish versus prosocial option. Indifference points were higher at social distance 50 and 100 in the autism group compared to the non-autistic group. Thus, autistic adults showed greater generosity to more socially distant others compared to non-autistic participants.

In a different study, Warnell et al. (2019) asked autistic adolescents and young adults to think of people at four different social distances with the closet (P1) being ‘the person with whom they were closest’ and the most socially distant person (P4) being ‘a person they had met personally, but did not know well at all’ (p. 871). At each social distance, participants started with a choice between $50 for themselves (self amount) and $100 for the other person (other amount). If participants chose $50 for themselves, then on the next trial participants chose between $25 for themselves and $100 for the other person. Conversely, if participants chose $100 for the other person on the first trial, then on the next trial, they chose between $75 for themselves and $100 for the other person. This sequential adjustment of the self-amount continued for six trials until an indifference point was reached: the subjective value equivalent to the other person receiving $100. The mean indifference point at P1 and P2 for the autistic group was lower compared to a matched non-autistic group. There were no group differences at P3 or P4. Thus, autistic participants were less generous to close others compared to the non-autistic group.

Both studies found flatter social discounting in autism: autistic participants distinguished less between people at different social distances. However, compared to non-autistic participants, in the work by Tei et al. (2019), this resulted in increased generosity to distance others, whereas in the work by Warnell et al. (2019), this resulted in reduced generosity to close others. A range of factors could account for their differing findings, such as the range of social distance levels (four vs six), the age of the participants or how social discounting was measured. Thus, given the discrepancy in these two studies, our first aim was to gather further data concerning the effect of autism on social discounting. Crucially, however, we also aimed to extend this previous work by investigating the role of framing effects on social discounting in autism.

Framing effects in autism

Social discounting is also influenced by how the resource allocation task is framed (Sellitto et al., 2021). Participants are more generous to socially distant others if their choices are framed as preventing someone losing money (loss frame) compared to when they result in that person gaining money (gain frame), despite the financial outcomes being objectively equivalent across conditions.

It is unknown whether, and to what extent, autistic individuals show framing effects in social discounting. Rozenkrantz et al. (2021) suggested autism is characterised by ‘enhanced rationality’ as autistic individuals, who have cognitive and language skills comparable to neurotypical participants, are less prone to cognitive biases. This view of autism appreciates that while autistic people face challenges in their daily lives, in certain domains, autistic individuals show fewer biases than their neurotypical peers. For example, autistic individuals are less susceptible to the optimism bias, so learn equally from desirable and undesirable information (Kuzmanovic et al., 2019), and are less vulnerable to sunk-costs, so decisions are less influenced by unrecoverable costs that have already been incurred (Fujino et al., 2019).

In terms of framing effects in autism, two studies have investigated self-benefitting financial decisions under risk where decisions where framed as either personal gains or losses (De Martino et al., 2008; Shah et al., 2016). Adult participants chose between a sure option (i.e. gain: keep £20 from a total of £50 vs loss: lose £30 from a total of £50) compared to a gamble option in which they could keep all of a monetary endowment (i.e. £50) or lose it all at some probability. Neurotypical participants gambled more in the loss frame compared to the gain frame despite the potential financial outcomes being equivalent across conditions (Kahneman & Tversky, 1979). Both studies showed that autistic and neurotypical participants showed a preference for gambling in the loss frame compared to the gain frame, but this effect was significantly reduced in the autism group. In addition, De Martino et al. (2008) found neurotypical participants had a stronger autonomic response, as measured by skin conductance, when decisions were framed as a loss compared to when they were framed as a gain. The skin conductance responses of autistic participants did not differ between the loss and gain frame suggesting that autistic individuals could incorporate emotional information differently during decision-making, which could help to explain enhanced rationality in autism. That said, Rozenkrantz et al. (2021) have proposed that a range of factors could influence enhanced rationality in autism, including differences in reward processing, intuition, attention to detail and sensitivity to contextual information.

It is also worth highlighting that autistic individuals do not show enhanced rationality across all domains. For example, in terms of temporal discounting, there is evidence to suggest that autistic adolescences show a heightened preference for smaller-sooner rewards (e.g. £60 now) over larger-later rewards (e.g. £100 in 1 year) compared to non-autistic participants (Carlisi et al., 2017; Chantiluke et al., 2014), although not all studies find these differences (e.g. Demurie et al., 2012; Warnell et al., 2019).

Current study

In this study, we aimed to extend previous work investigating social discounting in autism. Specifically, we aimed to determine the influence of framing effects on autistic participants’ generosity towards people at different social distances. While prior work on social discounting in autism is equivocal (cf. Tei et al., 2019; Warnell et al., 2019), previous studies have, as mentioned above, consistently demonstrated a reduced susceptibility to cognitive biases in autism (Rozenkrantz et al., 2021) with two studies showing a reduced sensitivity to framing effects (De Martino et al., 2008; Shah et al., 2016). Both these studies investigated the influence of framing effects on self-benefitting financial decisions under risk. Here we aimed to determine the influence of framing effects on prosocial monetary decisions towards individuals at different social distances.

We used an established measure of social discounting (Jones & Rachlin, 2006; Margittai et al., 2015; Schweda et al., 2020; Sellitto et al., 2021; Tei et al., 2019) in which the dependent variable was the amount of money participants were willing to forgo in favour of someone else’s benefit at each social distance. Moreover, we framed the decision as either preventing a loss or causing a gain for another person (Schweda et al., 2020; Sellitto et al., 2021). Specifically, participants always chose between a selfish option, where the participant received a higher pay off for themselves but the other recipient received a zero payoff, and a generous option, where the participant and the other recipient both received an equal non-zero payoff (75€ for both). The key difference between the frame conditions was that in the gain frame, if the participant chose the generous option this involved causing a 75€ gain for the other recipient, whereas in the loss frame the generous option involved preventing a loss of a 75€ endowment for the other recipient. Crucially, the payoff structure was mathematically equivalent across conditions so the potential money the participant and the other recipient could receive was identical in the loss and gain frame. With this design, we could determine the influence of social distance and frame, and their potential interaction, on generosity in both autistic and neurotypical participants.

Method

Participants

We recruited 53 participants aged between 20 and 47 years who were matched on age and gender across the experimental groups (autism: n = 28 (12 women; 43%), mean (SD) age = 30.61 (8.63) years; neurotypical: n = 25 (13 women; 52%), mean (SD) age = 30.28 (5.65); all ps > 0.69). All autistic participants had a formal diagnosis of autism (according to (International Classification of Diseases, 10th Revision (ICD-10): F84.0 or F84.5) provided by an experienced consultant psychiatrist based on a systematic assessment implemented at an outpatient clinic at the Department of General Psychiatry 2 at the University Hospital Düsseldorf. One autistic participant received their diagnosis as a child; all other participants (96.4%) received their diagnosis as adults.

A greater proportion of participants in the neurotypical group (19/25; 76%) reported having a university degree compared with those in the autism group (9/28; 32%; p = 0.001). Out of the 28 autistic participants, 23 provided information concerning their employment status: 15 reported being employed (65%), 5 (22%) were studying or in training and 3 (13%) indicated that they were studying, training nor employed. Specific data on socioeconomic status and race/ethnicity were not recorded.

Ethical approval was obtained for all procedures and were conducted according to the Declaration of Helsinki. Participants provided written informed consent before completing the study. After completing the experimental task, participants were asked to complete a series of online questionnaires and tasks, including the Autism Quotient (Baron-Cohen et al., 2001) and the extended version of the first part of the Culture Fair Intelligence Test (CFT 20-R; Grundintelligenztest Skala 2; Weiß, 2006). AQ scores were obtained from 45 participants (autism = 28/28; neurotypical = 17/25; missing data were due to a technical error during data collection) and CFT 20-R scores, measuring non-verbal IQ, were obtained from 44 participants (autism = 27/28; neurotypical = 17/25). AQ scores were higher in the autism group (mean (SD) autism = 40.57 (6.63) neurotypical = 12.65 (4.15), p < 0.001) and the autism group had higher non-verbal IQ scores (autism = 122.26 (15.87) neurotypical = 108.53 (17.10), p = 0.012). We conducted additional analysis to ensure that our reported effects were not due to the IQ differences between the groups (see Supplemental Table S3). We collected several additional trait variables. For a full comparison of the psychological traits and abilities of the two groups, see Supplemental Table S1. The study was not preregistered.

Social discounting task

Before the task, participants were told that social distance refers to how close a person is to them and were presented with a scale from 1 to 100 showing themselves presented by a purple figure on the far left. A yellow figure was then shown at various social distances along this scale: 1, 5, 10, 20, 50 and 100. For social distances 1, 5, 10 and 20, participants were asked to think of someone they know at these social distances and to write the first name of this person and their relationship to them. For social distance 50, participants were told to think of someone they may have seen a few times but do not really know. For social distance 100, they were told to think of a stranger whom they do not know and will never see again. For the other social distances, participants were not given explicit instructions concerning who to think of. However, participants were instructed not to think of anyone they have negative feelings towards and not anyone with whom they ‘share a household’ – either by actually living with them or by means of a joint bank account.

In the task, participants decided how to distribute money between themselves and the people at the different social distances. Participants were informed that two participants would be randomly selected, and for each of these participants, one of their decisions would be randomly selected and paid out. Thus, decisions were not hypothetical but rather fully incentive-compatible in that participants knew their decisions could have actual financial consequences.

On each trial in the gain frame (Figure 1(a)), participants were told that the other person has 0€ and they must make a choice between 75€ for both themselves and the other person at one of the social distance levels 1, 5, 10, 20, 50 or 100 (generous option) and nothing for the other person and a potentially larger amount (75€, 85€, 95€, 105€, 115€, 125€, 135€, 145€ or 155€) for themselves (selfish option). In the loss frame (Figure 1(b)), participants were told that the other person has 75€ and they must make a choice between 75€ for themselves and nothing for the other person (generous option), or the loss of 75€ for the other person and a larger amount (75€, 85€, 95€, 105€, 115€, 125€, 135€, 145€ or 155€) for themselves (selfish option). Note that the participants made 108 choices (six social distances × nine different selfish amounts in both the loss and gain frame) which were presented in a randomised order. Participants were informed that the potential recipient was unaware of both their initial endowment (75€ in the loss frame or 0€ in the gain frame). Instead, potential recipients would only be aware of the final payoff when the participant’s choice was implemented.

Figure 1.

An example of a trial in the (a) gain and (b) loss frame. (a) In the gain frame, a generous choice was described as a gain of 75€ for the other person; (b) in the loss frame, a generous choice was described as preventing the loss of 75€ for the other person. The objective economic outcomes were identical across frames: participants chose between 75€ for both themselves and the other person, or more money for themselves (e.g. 115€) and 0€ for the other person. The social distance of other person varied on each trial.

In sum, the gain and loss frames were mathematically equivalent so yielded identical potential payoffs across conditions both for the participant themselves and the recipient. For example, in Figure 1, in both frames a selfish choice yields 0€ for the other person and 115€ for the participants themselves, whereas a generous choice results in 75€ for the participant and 75€ for the other person. However, the fundamental difference between the conditions was that when the other participant received €0, this was framed as a loss of an initial endowment in the loss frame compared to a zero payoff in the gain frame.

Analysing social discounting

We used both linear mixed effects models and an established modelling approach to determine the extent of social discounting in the loss frame and gain frame (Margittai et al., 2015, 2018; Strombach et al., 2016). We used the lme4 package in R (Bates et al., 2015; R Core Team, 2020) to conduct linear mixed effects models.

Money forgone

The dependent variable was the amount of money participants decided to forgo on each trial. As the generous option was always to keep 75€ for oneself and allow the other person to receive (gain frame) or keep (loss frame) 75€, the potential amount forgone was the selfish amount on offer on each trial minus 75€. For example, if the participant could receive 115€ (and 0€ for the other person) but chose the generous option (75€ for both), then they decided to forgo 40€ on this trial. If the participants chose the selfish option, then the amount forgone was 0€.

Model predictors

We included the predictors ‘group’ (autism vs neurotypical), ‘frame’ (loss vs gain) and ‘rank social distance’ to linearise the social distance function as the gaps between the social distances were not incremental (i.e. 1, 5, 10, 20, 50, 100; Schweda et al., 2020). The models included random intercepts for participants and random slopes for within-subject factors: ‘frame’ and ‘rank social distance’ (Barr et al., 2013; Matuschek et al., 2017). We then compared the full model containing the three-way interaction between the predictors to a simpler model containing only the two-way interactions. The more complex model (Akaike information criterion (AIC) = 52,321; Bayesian information criterion (BIC) = 52,420) revealed no significant three-way interaction between group, frame and rank social distance and did not provide a better fit to the data than the simpler model (AIC = 52,319; BIC = 52,412; p = 0.836). Thus, we report the results from the simpler model.

Modelling social discounting

For the modelling approach, for each participant, we calculated the indifference point at each social distance separately in the gain and loss frame using a logistic regression – the amount at which the probability of choosing the selfish or generous amount was 50%. If participants made only selfish choices, then the indifference point was set to 75; conversely, if they only made generous choices, then it was set to 160. Next, the indifference points for each social distance were fitted to an established hyperbolic model (Jones & Rachlin, 2006)

v = \frac{V}{1 + k D}

Here, v represents the value of the generosity, V can be interpreted as the generosity shown to close others (it is the intercept at social distance 0), D is the social distance, and, crucially, k is the social discounting parameter that describes the extent to which generosity declines across social distance.

Area under the curve

In addition, we calculated the area under the curve (AUC) which is a model-free measure of the total discount rate independent of any assumptions concerning the functional form of discounting (Myerson et al., 2001). To calculate AUC for each participant, we used the mean amount of money forgone at each social distance in each frame and then followed the approach of Myerson et al. (2001), calculating the sum of trapezoids between successive social distances. AUC was expressed as a value between 0 and 1 so was proportional to the maximum AUC. For example, for participants who always chose the generous option, AUC was 1.

Community involvement

Autistic individuals or the wider autism community were not involved in the development, design, implementation or interpretation of this research.

Results

The data and code association with the analyses are available on the Open Science Framework https://osf.io/89rjq/.

Enhanced generosity and reduced framing effects in autism

The linear mixed effects model revealed a main effect of rank social distance on amount forgone (estimate = −6.44, SE = 0.575, p < 0.001), indicating a decline in the amount of money forgone at increasing social distance. There was also a main effect of group (estimate = −4.72, SE = 2.20, p = 0.037) with the autism group foregoing more money overall (M = 30.0, SE = 1.75) compared to the neurotypical group (M = 25.5, SE = 1.85).

There was a significant interaction between rank social distance and group across both frame conditions (estimate = 3.29, SE = 0.749, p < 0.001). Simple slope analysis using the R package interactions (Long & Long, 2022) revealed that autistic participants showed a flatter social discounting slope (estimate = −3.15, SE = 0.549, p < 0.001) compared to neurotypical participants (estimate = −6.44, SE = 0.575, p < 0.001) (Figure 2(a)). Post hoc t tests revealed that compared to the neurotypical group, the mean amount forgone in the autism group was significantly greater at further social distances (i.e. social distance 100; p = 0.002; see Supplemental Table S2).

Figure 2.

Panels a and b show the mean amount forgone in the neurotypical group and the autism group at each social distance (a) and in the loss frame and gain frame (b). The solid lines and ribbons in panel a are the model estimates for the interaction effects between group and rank social distance plus their respective 95% confidence intervals. Post hoc t tests revealed a significant group difference at rank social distance six. In panel b, the lines represent the model estimates (±standard errors of the mean) for the interaction effects between group and frame. Post hoc t tests revealed a significant group difference in the gain frame. The means for each participant in the autism group and neurotypical group are displayed by the red crosses and blue squares, respectively. Panels c, d and e are boxplots showing the distribution of k, V and AUC, respectively. V represents the generosity shown to close others, k quantifies the decline in generosity across social distance (higher k indicates more social discounting) and AUC is a summary measure of the total generosity shown across the social distances. As above, the individual data points for each participant are displayed by the red crosses and blue squares.

There was a significant interaction between frame and group (estimate = −4.46, SE = 2.15, p = 0.043) driven by a larger framing effect in the neurotypical group (mean (SE): gain = 21.9 (1.90), loss = 29.1 (2.12) compared to the autism group (gain = 28.7 (1.80), loss = 31.4 (2.00) (Figure 2(b)). Post hoc contrasts revealed a significant difference between the groups in the mean amount forgone in the gain frame (p = 0.019; Bonferroni-corrected) but not in the loss frame (p = 0.849). There was also a significant interaction between rank social distance and frame (estimate = 1.81, SE = 0.373, p < 0.001) with flatter social discounting slope in the loss frame (estimate = −4.64, SE = 0.575, p < 0.001) compared to the gain frame (estimate = −6.44, SE = 0.575, p < 0.001).

As the parameters from the modelling and AUC values were not normally distributed (see Figure 2(c) to (e)), we used Mann–Whitney U tests using the wilcox.test function in R to calculate differences between the groups. The p values were Bonferroni-corrected to control for multiple testing (i.e. in the loss and gain frame). Due to missed trials, parameters and AUC could not be calculated for one autistic participant in the gain frame.

Social discounting, represented by k, quantifies the decline in generosity across social distance. This parameter was smaller in the autism group compared to the neurotypical group in both the gain frame (medians (lower quartile, upper quartile): neurotypical = 0.0068 (0.0041, 0.0133); autism = 0.0002 (0.0000, 0.0044); U = 510, p = 0.003) and loss frame (neurotypical = 0.0029 (0.0000, 0.0086); autism = 0.0000 (0.0000, 0.0006); U = 546.5, p < 0.001) (Figure 2(c)). Thus, autistic participants showed a less marked decline in generosity at increasing social distance. Generosity to close others, represented by V, was greater in the neurotypical group compared to the autistic group in the loss frame (neurotypical = 85.00 (79.89, 89.07); autism = 85.00 (73.57, 85.00); U = 486.5, p = 0.029) but did not differ between the groups in the gain frame (medians: neurotypical = 79.81 (69.50, 85.84); autism = 85.00 (71.20, 85.60); U = 339, p = 1.00) (Figure 2(d)). Finally, AUC, a summary measure of the total generosity shown across the social distances, was greater in the gain frame in the autistic group compared to the neurotypical group (neurotypical = 0.262 (0.178, 0.444); autism = 0.769 (0.312, 0.981); U = 172, p = 0.005) but did not differ between the groups in the loss frame (neurotypical = 0.586 (0.263, 0.944). autism = 0.976 (0.458, 1.000); U = 241, p = 0.102) (Figure 2(e)).

To ensure these differences were not due to IQ differences between the groups (see Methods), we conducted a series of robust linear regressions including IQ score and group as predictors. These analyses showed that group differences in k gain, k loss and AUC gain remained even when controlling for IQ scores. Group differences in V loss were no longer significant (see Supplemental Table S3).

Discussion

We investigated social discounting in autism and report several key findings. First, compared to a matched neurotypical group, autistic participants were more generous – they were more willing to forgo money for another person’s benefit. Second, social discounting was reduced in autism – the amount of money that autistic participants were willing to forgo for someone else’s benefit varied less across social distances resulting in increased generosity to socially distant others. Finally, compared to the neurotypical participants, autistic participants were less susceptible to framing effects when deciding whether to forgo money for someone else’s monetary gain compared to when making the same decision to prevent someone’s monetary losses.

Greater generosity to more socially distant others in autism

Our findings support the results from two previous studies showing that autistic participants are less influenced by social distance manipulations (Tei et al., 2019; Warnell et al., 2019). Our data support Tei et al.’s (2019) findings showing increased generosity to socially distant others in autism. We did not find any differences in the amount of money forgone at close social distances (Figure 2(a)) which is not consistent with Warnell et al.’s (2019) findings who found reduced generosity to closer others in autism. That said, our modelling approach revealed smaller V values in the loss frame in the autism group compared to neurotypicals, indicative of reduced generosity to closer others (Figure 2(d)). However, this effect was driven by a flatter (or completely flat) social discounting function in autistic participants who showed very small social discounting parameters (k) in the loss domain (Figure 2(c)). Moreover, when we controlled for differences in IQ scores, group differences in V loss were no longer significant (see Supplemental Table S3).

As in the work by Tei et al. (2019), we only investigated adults and had six social distances including a stranger condition. We also incentivised participants by emphasising that some of their choices could be randomly selected and actually paid out. In contrast, Warnell et al. (2019) only included four social distances (the furthest distance being a person the participant had personally met but did not know well at all) and asked younger adults and adolescents to make hypothetical choices. While all these factors could explain the different findings between the studies, we focus on two factors which could account our results: (1) autistic adults implemented a fairness norm more consistently across social distances; (2) autistic participants showed a reduced sensitivity to social information.

The experimental paradigm we used to measure social discounting was very similar to that used by Tei et al. (2019) in that the generous choice always involved an equal share of the money (75€ for both the participant and other recipient), whereas the selfish option involved a greater amount for the participant themselves and nothing for the other recipient (Strombach et al., 2016). Contrast this to Warnell et al.’s (2019) titration approach, where participants started with a choice between $50 for themselves and $100 for the other recipient and could then progressively keep more or less money for themselves depending on their previous choices. Thus, in our study and that of Tei et al. (2019), but not in Warnell et al. (2019), a participant preferring an equal share of resources between themselves and the other person would be inclined towards the generous option, resulting in more generous choices. Thus, reduced social discounting in autism could be due to a more consistent implementation of a fairness norm, that is, choosing an equal split of the money.

In a recent meta-analysis, Ryan-Enright et al. (2022) did not find consistent differences in rates of sharing in autistic children but did find that fairness and equity norms were implemented less consistently (Paulus & Rosal-Grifoll, 2017; Schmitz et al., 2015). Thus, while these norms may develop differently in autistic children, possibly due to differences in theory of mind (Sally & Hill, 2006), we propose that once these fairness norms are learned and established, they are implemented more consistently when given the chance to share resources in later adolescence and adulthood (Ikuse et al., 2018; Tei et al., 2019). For example, autistic children, but not adolescents, were more likely to accept unfair offers in the ultimatum game (Jin et al., 2020). This view is consistent with findings from other domains in which some autistic individuals use rules as part of compensatory strategies to navigate their social environments (Livingston et al., 2020). However, little is known about how exactly this occurs and what the benefits and costs are for the individual. Here, longitudinal studies are required to investigate relevant developmental trajectories.

In addition, we argue that a reduced sensitivity to social information when sharing money in autistic adults could also underly our effects. Whereas neurotypical individuals give more money when being observed by others, donation behaviour in autism is less influenced by audience effects (the increase in generosity usually shown by neurotypical participants when being observed by others) potentially due to differences in reputation management (Cage et al., 2013; Frith & Frith, 2011; Izuma et al., 2011). Moreover, Ikuse et al. (2018) found that autistic adults distributed more money in an ultimatum game and this greater generosity was positively associated with AQ scores. Crucially, while neurotypical adults adjusted the amount depending on the social relevance of the visual background of the experimental task, giving more when pictures of eyes were shown compared to a neutral background or images of flowers, autistic adults tended to distribute around ¥100, so roughly half of the original endowment of ¥200 yen, regardless of the background (Ikuse et al., 2018). In other words, as in this study, autistic adults preferred an equal split of the money with other people independent of biasing information (e.g. social distance, framing, observation by others and visual background), resulting in greater generosity. Together, our findings suggest that enhanced generosity towards more socially distant others in autism is related to both more consistent responding, potentially related to a more consistent implementation of a fairness norm, and a reduced sensitivity to social information and how one may be perceived by others. In other words, both the social and non-social aspects of autism could have contributed to our effects.

A related possibility is that the relative subjective closeness participants felt towards individuals at different social distances was not the same in autistic participants and the neurotypical control group. Thus, larger differences in perceived closeness between individuals at close and more distant social distances in neurotypical participants compared to autistic participants could to some extent explain the effects. For example, it is possible that autistic individuals feel subjectively closer to strangers compared to neurotypical participants. Mazurek (2014) found that 40% of autistic adults did not have a close or best friend. Hence, it is also plausible to assume that neurotypical controls attach higher value to socially close others than autistic participants, hence explaining the steeper discounting in neurotypicals.

Reduced framing effects in autism

Autistic participants were less influenced by the framing manipulation. Previous studies have shown that neurotypical participants are more generous when their decisions prevent losses in others compared to when they result in gains for others (Schweda et al., 2020; Sellitto et al., 2021). Although autistic participants were also more generous in the loss frame compared to the gain frame, this effect was reduced compared to neurotypical participants. Previous studies in autism have focused on framing effects in the context of self-benefitting decisions (De Martino et al., 2008; Shah et al., 2016). In these studies, participants had the choice between some proportion of a monetary endowment, which was framed as either a certain gain or a certain loss, and a gamble in which they had the chance to keep the whole endowment or risk losing it all. Our results extend these findings in several important ways. First, our task did not involve a decision under risk. Thus, our findings support the view that differences in framing effects in autism are not simply the result of differences in risk aversion or how probabilities are calculated, but result from a reduced susceptibility to cognitive biases (De Martino et al., 2008). Second, decisions in our study were not purely self-benefitting as participants choices also had potential consequences for another person. This demonstrates that a reduced aversion to incurring losses relative to receiving gains in autism generalises to contexts in which the potential beneficiary of these choices is another person.

Several explanations have been put forward to explain reduced framing effects in autism. De Martino et al. (2008) interpreted their findings in terms of a ‘two-systems’ model (Evans, 2003) or ‘dual process’ theory (Brosnan et al., 2016): autistic individuals may be more reliant on deliberative, analytic decision-making processes at the expense of intuition, potentially related to increased attention to detail. However, in a recent series of studies with large sample sizes, Taylor et al. (2022) found no differences in ‘intuitive’ versus ‘deliberative’ decision-making in autism. Thus, future studies will need to determine exactly which aspects of autism account for a reduced susceptibility to framing effects.

Implications and future directions

Our study and those of others (Ikuse et al., 2018; Tei et al., 2019), which show increased generosity in autism, fit with an emerging view that the particular pattern of strengths and difficulties autistic individuals show in terms of their cognition and behaviour can have positive and negative consequences both for the individual themselves and those around them (Rozenkrantz et al., 2021). The specific circumstances under which autistic participants display either increases or decreases in prosocial behaviour is an exciting avenue for future research.

Autism and autistic traits have historically been linked to reductions in prosocial behaviour, for example, via empathy differences (Zhao et al., 2019). However, just as autistic people do not have a pervasive empathy ‘deficit’ (for a discussion of the harm this mischaracterisation has caused autistic people, see Fletcher-Watson & Bird, 2020), our data paint a more complex picture of prosocial behaviour in autism. A greater understanding and awareness that autism can result in both increases and decreases in prosocial behaviour will hopefully help to change public perceptions of autism and improve opportunities for autistic people. For example, higher autistic traits have been linked to greater accuracy in predicting social phenomena, such as group think and social loafing (Gollwitzer et al., 2019), and autistic individuals can show broader and more diverse empathic responses, for example, in terms of object personification (White & Remington, 2019). Furthermore, it has been increasingly recognised that social interaction difficulties associated with a range of conditions can only be adequately characterised by considering all interaction partners, their potentially diverse cognitive styles and perceptual abilities as well as the degree of interpersonal matching and similarity (Bolis et al., 2021; Schilbach, 2016).

Finally, while we replicated previous studies in terms of social discounting effects (Tei et al., 2019) and framing effects (De Martino et al., 2008; Shah et al., 2016) in autism, our sample was relatively small (28 autistic participants, 25 neurotypical participants) and we did not preregister the study. Future preregistered studies should aim to replicate our findings in larger and more diverse samples of autistic participants to test the robustness of the observed effects.

Conclusion

We show that compared to a neurotypical group, autistic adults were more generous to other people, which was driven by a greater generosity to more socially distant others. We propose that this increased generosity to strangers is driven by autistic adults implementing fairness norms more consistently and differences in sensitivity to social information. We also build on previous work by showing that reduced framing effects in autism also extend to prosocial decision-making. Together, our results support the view of ‘enhanced rationality’ in autism as participants’ prosocial decision-making was less influenced by potential biasing information, such as the closeness of the potential recipient or how the choice was framed. More generally, our results challenge historical accounts of autism by showing that the differences seen in autism can also have prosocial consequences.

Supplemental Material

sj-pdf-1-aut-10.1177_13623613231190674 – Supplemental material for Autistic adults show enhanced generosity to socially distant others

Supplemental material, sj-pdf-1-aut-10.1177_13623613231190674 for Autistic adults show enhanced generosity to socially distant others by Paul AG Forbes, Irini Chaliani, Leonhard Schilbach and Tobias Kalenscher in Autism

Footnotes

Acknowledgements

We thank Katharina Waldthausen, Paulina Mondovits, Fabiana Lohr, Nora Bucher and Jana Totzek for their help with data collection and Damon Dashti and Philipp Stass for help with data checking.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This work was supported by a grant from the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG, grant no. KA 2675/7-1 to TK).

ORCID iD

Tobias Kalenscher

Supplemental material

Supplemental material for this article is available online.

References

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders: DSM-5 (Vol. 10). American Psychiatric Association.

Archambault

Kalenscher

de Laat

(2020). Generosity and livelihoods: Dictator game evidence on the multidimensional nature of sharing among the Kenyan Maasai. Journal of Behavioral Decision Making, 33(2), 196–207.

Baron-Cohen

Wheelwright

Skinner

Martin

Clubley

(2001). The autism-spectrum quotient (AQ): Evidence from Asperger syndrome/high-functioning autism, males and females, scientists and mathematicians. Journal of Autism and Developmental Disorders, 31(1), 5–17.

Barr

D. J.

Levy

Scheepers

Tily

H. J.

(2013). Random effects structure for confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language, 68(3), 255–278.

Bates

Mächler

Bolker

Walker

. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1), 1–48. https://doi.org/10.18637/jss.v067.i01

Bolis

Lahnakoski

J. M.

Seidel

Tamm

Schilbach

(2021). Interpersonal similarity of autistic traits predicts friendship quality. Social Cognitive and Affective Neuroscience, 16(1–2), 222–231.

Brosnan

Lewton

Ashwin

(2016). Reasoning on the autism spectrum: A dual process theory account. Journal of Autism and Developmental Disorders, 46(6), 2115–2125.

Cage

Pellicano

Shah

Bird

(2013). Reputation management: Evidence for ability but reduced propensity in autism. Autism Research, 6(5), 433–442.

Carlisi

C. O.

Norman

Murphy

C. M.

Christakou

Chantiluke

Giampietro

Simmons

Brammer

Murphy

D. G.

, MRC AIMS consortium, Mataix-Cols

Rubia

(2017). Comparison of neural substrates of temporal discounting between youth with autism spectrum disorder and with obsessive-compulsive disorder. Psychological Medicine, 47(14), 2513–2527. https://doi.org/10.1017/S0033291717001088

10.

Chantiluke

Christakou

Murphy

C. M.

Giampietro

Daly

E. M.

Ecker

Brammer

Murphy

D. G.

Rubia

(2014). Disorder-specific functional abnormalities during temporal discounting in youth with Attention Deficit Hyperactivity Disorder (ADHD), Autism and comorbid ADHD and Autism. Psychiatry Research: Neuroimaging, 223(2), 113–120. https://doi.org/10.1016/j.pscychresns.2014.04.006

11.

Chevallier

Kohls

Troiani

Brodkin

E. S.

Schultz

R. T.

(2012). The social motivation theory of autism. Trends in Cognitive Sciences, 16(4), 231–239.

12.

De Martino

Harrison

N. A.

Knafo

Bird

Dolan

R. J.

(2008). Explaining Enhanced Logical Consistency during Decision Making in Autism. Journal of Neuroscience, 28(42), 10746–10750. https://doi.org/10.1523/JNEUROSCI.2895-08.2008

13.

Demurie

Roeyers

Baeyens

Sonuga-Barke

(2012). Temporal discounting of monetary rewards in children and adolescents with ADHD and autism spectrum disorders. Developmental Science, 15(6), 791–800. https://doi.org/10.1111/j.1467-7687.2012.01178.x

14.

Evans

J. S. B.

(2003). In two minds: Dual-process accounts of reasoning. Trends in Cognitive Sciences, 7(10), 454–459.

15.

Fletcher-Watson

Bird

(2020). Autism and empathy: What are the real links? Autism, 24(1), 3–6.

16.

Frith

(2011). Reputation management: In autism, generosity is its own reward. Current Biology, 21(24), R994–R995.

17.

Fujino

Tei

Itahashi

Aoki

Ohta

Kanai

Kubota

Hashimoto

Nakamura

Kato

Takahashi

(2019). Sunk cost effect in individuals with autism spectrum disorder. Journal of Autism and Developmental Disorders, 49(1), 1–10. https://doi.org/10.1007/s10803-018-3679-6

18.

Gollwitzer

Martel

McPartland

J. C.

Bargh

J. A.

(2019). Autism spectrum traits predict higher social psychological skill. Proceedings of the National Academy of Sciences, 116(39), 19245–19247.

19.

Ikuse

Tani

Itahashi

Yamada

Ohta

Morita

Arai

Saga

Tokumasu

Ohta

Sato

Iwanami

(2018). The effect of visual cues on performance in the ultimatum game in individuals with autism spectrum disorder. Psychiatry Research, 259, 176–183. https://doi.org/10.1016/j.psychres.2017.10.013

20.

Ito

Saeki

Green

(2011). Sharing, discounting, and selfishness: A Japanese-American comparison. The Psychological Record, 61(1), 59–75.

21.

Izuma

Matsumoto

Camerer

C. F.

Adolphs

(2011). Insensitivity to social reputation in autism. Proceedings of the National Academy of Sciences, 108(42), 17302–17307.

22.

Jin

Wang

Xiao

Qiu

Weng

Fang

(2020). The fair decision-making of children and adolescents with high-functioning autism spectrum disorder from the perspective of dual-process theories. BMC Psychiatry, 20(1), 1–11.

23.

Jones

Rachlin

(2006). Social discounting. Psychological Science, 17(4), 283–286. https://doi.org/10.1111/j.1467-9280.2006.01699.x

24.

Jones

B. A.

(2022). A review of social discounting: The impact of social distance on altruism. The Psychological Record, 72(3), 511–515. https://doi.org/10.1007/s40732-021-00488-5

25.

Kahneman

Tversky

(1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263–291.

26.

Kuzmanovic

Rigoux

Vogeley

(2019). Brief report: Reduced optimism bias in self-referential belief updating in high-functioning autism. Journal of Autism and Developmental Disorders, 49(7), 2990–2998. https://doi.org/10.1007/s10803-016-2940-0

27.

Livingston

L. A.

Shah

Milner

Happé

(2020). Quantifying compensatory strategies in adults with and without diagnosed autism. Molecular Autism, 11(1), 1–10.

28.

Long

J. A

. (2022). Interactions: Comprehensive, user-friendly toolkit for probing interactions. R package (Version 1.1.6). https://cran.r-project.org/package=interactions

29.

Margittai

Strombach

Van Wingerden

Joëls

Schwabe

Kalenscher

(2015). A friend in need: Time-dependent effects of stress on social discounting in men. Hormones and Behavior, 73, 75–82.

30.

Margittai

van Wingerden

Schnitzler

Joëls

Kalenscher

(2018). Dissociable roles of glucocorticoid and noradrenergic activation on social discounting. Psychoneuroendocrinology, 90, 22–28. https://doi.org/10.1016/j.psyneuen.2018.01.015

31.

Matuschek

Kliegl

Vasishth

Baayen

Bates

(2017). Balancing Type I error and power in linear mixed models. Journal of Memory and Language, 94, 305–315.

32.

Mazurek

M. O.

(2014). Loneliness, friendship, and well-being in adults with autism spectrum disorders. Autism, 18(3), 223–232. https://doi.org/10.1177/1362361312474121

33.

Myerson

Green

Warusawitharana

(2001). Area under the curve as a measure of discounting. Journal of the Experimental Analysis of Behavior, 76(2), 235–243. https://doi.org/10.1901/jeab.2001.76-235

34.

Nowak

M. A

. (2006). Five rules for the evolution of cooperation. Science, 314(5805), 1560–1563.

35.

Olson

E. A.

Rosso

I. M.

Demers

L. A.

Divatia

Killgore

W. D. S.

(2016). Sex differences in psychological factors associated with social discounting. Journal of Behavioral Decision Making, 29(1), 60–66. https://doi.org/10.1002/bdm.1876

36.

Paulus

Rosal-Grifoll

(2017). Helping and sharing in preschool children with autism. Experimental Brain Research, 235(7), 2081–2088. https://doi.org/10.1007/s00221-017-4947-y

37.

Pellicano

(2012). The development of executive function in autism. Autism Research and Treatment, 2012, 1–8. https://doi.org/10.1155/2012/146132

38.

R Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/

39.

Rozenkrantz

D’Mello

A. M.

Gabrieli

J. D. E.

(2021). Enhanced rationality in autism spectrum disorder. Trends in Cognitive Sciences, 25(8), 685–696. https://doi.org/10.1016/j.tics.2021.05.004

40.

Ryan-Enright

O’Connor

Bramham

Taylor

L. K.

(2022). A systematic review of autistic children’s prosocial behaviour. Research in Autism Spectrum Disorders, 98, Article 102023. https://doi.org/10.1016/j.rasd.2022.102023

41.

Sally

Hill

(2006). The development of interpersonal strategy: Autism, theory-of-mind, cooperation and fairness. Journal of Economic Psychology, 27(1), 73–97.

42.

Schilbach

(2016). Towards a second-person neuropsychiatry. Philosophical Transactions of the Royal Society B: Biological Sciences, 371(1686), Article 20150081.

43.

Schmitz

E. A.

Banerjee

Pouw

L. B.

Stockmann

Rieffe

(2015). Better to be equal? Challenges to equality for cognitively able children with autism spectrum disorders in a social decision game. Autism, 19(2), 178–186.

44.

Schweda

Margittai

Kalenscher

(2020). Acute stress counteracts framing-induced generosity boosts in social discounting in young healthy men. Psychoneuroendocrinology, 121, Article 104860. https://doi.org/10.1016/j.psyneuen.2020.104860

45.

Sellitto

Neufang

Schweda

Weber

Kalenscher

(2021). Arbitration between insula and temporoparietal junction subserves framing-induced boosts in generosity during social discounting. NeuroImage, 238, Article 118211. https://doi.org/10.1016/j.neuroimage.2021.118211

46.

Shah

Catmur

Bird

(2016). Emotional decision-making in autism spectrum disorder: The roles of interoception and alexithymia. Molecular Autism, 7(1), 43. https://doi.org/10.1186/s13229-016-0104-x

47.

Strombach

Jin

Weber

Kenning

Shen

Kalenscher

(2014). Charity begins at home: Cultural differences in social discounting and generosity. Journal of Behavioral Decision Making, 27(3), 235–245.

48.

Strombach

Margittai

Gorczyca

Kalenscher

(2016). Gender-specific effects of cognitive load on social discounting. PLOS ONE, 11(10), Article e0165289. https://doi.org/10.1371/journal.pone.0165289

49.

Taylor

E. C.

Farmer

G. D.

Livingston

L. A.

Callan

M. J.

Shah

(2022). Rethinking fast and slow processing in autism. Journal of Psychopathology and Clinical Science, 131(4), 392–406. https://doi.org/10.1037/abn0000734

50.

Tei

Fujino

Itahashi

Aoki

Ohta

Kubota

Hashimoto

Nakamura

Kato

Takahashi

(2019). Egocentric biases and atypical generosity in autistic individuals. Autism Research, 12(11), 1598–1608. https://doi.org/10.1002/aur.2130

51.

Vekaria

K. M.

Brethel-Haurwitz

K. M.

Cardinale

E. M.

Stoycos

S. A.

Marsh

A. A.

(2017). Social discounting and distance perceptions in costly altruism. Nature Human Behaviour, 1(5), Article 5. https://doi.org/10.1038/s41562-017-0100

52.

Warnell

K. R.

Maniscalco

Baker

Redcay

(2019). Social and delay discounting in autism spectrum disorder. Autism Research, 12(6), 870–877. https://doi.org/10.1002/aur.2085

53.

Weiß

R. H.

(2006). CFT 20-R: grundintelligenztest skala 2-revision. Hogrefe.

54.

White

R. C.

Remington

(2019). Object personification in autism: This paper will be very sad if you don’t read it. Autism, 23(4), 1042–1045.

55.

Zhao

Song

Shi

(2019). Autistic traits and prosocial behaviour in the general population: Test of the mediating effects of trait empathy and state empathic concern. Journal of Autism and Developmental Disorders, 49(10), 3925–3938. https://doi.org/10.1007/s10803-018-3745-0

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.10 MB