Abstract: Whether you believe free will exists has profound effects on your behaviour, across different levels of processing, from simple motor action to social cognition. It is therefore important to understand which specific lay theories are held in the general public and why. Past research largely focused on investigating free will beliefs (FWB, ‘Do you think free will exists?’), but largely ignored a second key aspect: free will attitudes (FWA, ‘Do you like/value will?’). Attitudes are often independently predictive of behaviour, relative to beliefs, yet we currently know very little about FWAs in the general public. One key issue is whether such attitudes are subject to biased, socially desirable responding. The vast majority of the general public strongly believes in the existence of free will, which might create cultural pressure to value free will positively as well. In this registered report, we used a very large (N = 1100), open available dataset measuring implicit and explicit attitudes towards free will and determinism to address this issue. Our results indicate that both explicit and implicit attitudes towards free will are more positive than attitudes towards determinism. We also show that people experience cultural pressure to value free will, and to devalue determinism. Yet, we found no strong evidence that this cultural pressure affected either implicit or explicit attitudes in this dataset.
4. Discussion
In this registered report, we investigated attitudes towards free will and determinism in a large lay sample for the first time. Using data from the AIID project, we analysed both explicit and implicit attitudes towards free will and determinism. Attitudes towards free will were generally positive, while attitudes towards determinism were generally negative. This finding was robust to changes in how explicit and implicit attitudes were computed from the data. We further showed that people experience cultural pressure to value free will, and to devalue determinism. Contrary to our hypotheses, this pressure did not seem to affect free will attitudes.
4.1. Mapping free will attitudes in the general public
There is much prior research demonstrating that free will beliefs are dynamic, i.e. that they can be manipulated experimentally [7,16,21,37]. Such experimental manipulations often lead only to small effects, and there is little research directly addressing why changing free will beliefs is so difficult. Furthermore, free will belief manipulations probably affect different people to a varying degree, and we currently do not know which variables make a person susceptible to such manipulations. One factor that might partly explain these open questions is people's attitudes towards free will. A person with highly positive attitudes towards free will is likely to be less affected by FWB manipulations than a person with more negative attitudes towards free will. In order to test this hypothesis, we first need to map out and understand FWAs in the general public, which has not been done to date, however.
Our results show that both implicit and explicit attitudes towards free will are strongly positive in the general public. Only 11.39% of the participants explicitly valued free will negatively, while 34% chose the most positive available attitude rating on the scale, which mirrors similar findings for FWBs [4,5]. Furthermore, a lower proportion of participants explicitly valued determinism positively (35.96%), which again mirrors weaker belief in determinism in previous studies [5]. Interestingly, the mode of the explicit attitude distribution was nevertheless 0, indicating that many participants indicated to value free will and determinism to an equal degree. The same was not true for the implicit attitude distribution, which had a clearly positive mode instead. Thus, although many people indicated that they valued free will and determinism to an equal degree, implicit measures suggested that they in fact valued free will more strongly.
Additionally, our data demonstrate that there is considerable variance in FWAs (both explicit and implicit). Taken together, these results suggest that FWAs are a potential mediator for FWB manipulations, and might help explain why such manipulations are generally weak. Of course, these conclusions are tentative at the moment and will have to be confirmed empirically. Specifically, further research will need to directly assess the relation of FWBs and FWAs, and whether FWAs act as a mediator for FWB manipulations.
4.2. Cultural pressure and free will attitudes
One untested assumption in the free will belief literature is that responses to free will questionnaires are unbiased by cultural pressure. Here, we showed that in the domain of free will attitudes, participants indeed report cultural pressure to value free will, as well as a cultural pressure to devalue determinism. We further show that cultural pressure is stronger on free will than it is on determinism. Given this fact, one might expect that this cultural pressure affects attitudes in some measurable way. To investigate this possibility, we tested if more self-reported pressure to value free will led to more positive FWAs and whether more self-reported pressure led to more diverging explicit and implicit FWAs, since we expected cultural pressure to affect explicit attitudes more strongly than implicit ones. Both of these analyses yielded no significant results, however, leading us to conclude that cultural pressure did not affect FWA measures in our sample.
Although unexpected, there are some potential explanations for this finding. One option is that the cultural pressure measure did not fully capture the desired construct (biased/pressured responding). In order to assess cultural pressure, we only used two items that were rather blunt (e.g. ‘There is cultural pressure to think positive things about free will.’). Biased responding is clearly a wider concept than just self-rated cultural pressure, and is often measured using whole scales like the Balanced Inventory of Desirable Responding (BIDR) [38], a 40-item scale that measures both ‘impression management’ and ‘self-deceptive enhancement’. Such measures are more nuanced and of higher psychometric quality than the two cultural pressure items used in our main analysis.
While the BIDR was also part of the AIID dataset, we chose to use the cultural pressure items because the available sample size (214 < n < 233) was much higher than for the BIDR (13 < n < 55), and because these items were directly related to free will and determinism, while the BIDR is a much more general trait measure. However, the BIDR impression management and self-deceptive enhancement scales were included in the exploratory analysis correlating FWAs with personality traits, and we assessed BIDR scores in additional non-preregistered analyses. Impression management, which is arguably closer to the cultural pressure items used in the main analysis, did not correlate with either explicit (r = 0.09, p = 0.77, n = 13) or implicit (r = 0.06, p = 0.69, n = 43) attitudes. Thus, even this alternative measure of ‘cultural pressure’ showed no significant results. In contrast, self-deceptive enhancement did correlate with explicit attitudes (r = 0.63, p = 0.02, n = 13), but not with implicit attitudes (r = −0.02, p = 0.86, n = 55). Thus, it might be that people who tend to hold exaggerated positive self-descriptions report more positive feelings towards free will, relative to determinism. This exploratory finding is based on a very small sample size though and should be interpreted with caution until is it replicated in an independent, larger sample. Clearly, more research on this is needed, but this already demonstrates that free will attitudes might still be biased, just not by self-rated cultural pressure.
4.3. Exploratory analyses
In additional registered exploratory analyses, we found that age was positively related to free will attitudes. Interestingly, this finding is not consistent with a recent report that free will beliefs do not change with age [5]. This points to an interesting dissociation between free will beliefs and free will attitudes with increasing age that should be addressed in the future.
4.4. Directions for future research
Despite the many interesting findings in the current dataset, there remain several key open questions that should be addressed. First, the methods for measuring implicit attitudes towards free will need to be further refined. While the IAT is a much used and well validated instrument, its design only allows the assessment of two concepts in direct opposition. The IAT treats free will and determinism as two endpoints on a single attitude scale: the more you value free will, the less you value determinism, and vice versa. We know from free will belief research, however, that believing in free will does not automatically translate into disbelieving determinism [5], in fact one can believe in both at the same time (compatibilism, [1]). Thus, it might be that people also value both free will and determinism positively, yet the IAT is ill-suited to detect this pattern of free will related attitudes. Additionally, we found IAT scores to be more strongly related to deliberate thoughts on free will, than they are to gut feelings on free will, a pattern of results that we did not expect. Lastly, the words used to describe free will (intention, freedom, choice), and determinism (fixed, destined, arranged) have many uses that have little to do with the philosophical concepts of either free will or determinism, and might not precisely capture them. For instance, ‘arranged’ is often used to describe arranging a meeting, which has little to do with physical determinism. Taken together, this suggests that future research should invest effort into developing alternative measures of implicit attitudes towards free will.
Second, methods for measuring explicit attitudes towards free will and determinism should also be refined, and the precise wording of the attitude items might need to be revised. Additional effort should be invested into the precise wording of the attitude items. The AIID project used a wide range of different explicit attitude items, and here we focused on valence items (‘How positive or negative do you feel towards free will/determinism?’). For research specifically focusing on free will, it might be that other items (e.g. ‘Having free will is important to me’) are better suited to capture free will attitudes. Additionally, in order to make explicit and implicit attitude measures comparable here, we computed a difference score from free will and determinism valence ratings (val_diff). While this is useful from a methodological perspective, this procedure suffers from some of the same issues as the free will IAT. Just as the IAT cannot detect people who value both free will and determinism positively, as some compatibilists might do, the val_diff items cannot do so either. In the future, explicit attitude items for free will and determinism should be investigated separately, to more easily identify compatibilist participants.
Third, future work should focus on investigating free will beliefs and attitudes simultaneously, in order to describe and explain their interactions. Here, we focused on attitudes only, but in order to understand belief attitude interaction, e.g. whether free will attitudes mediate effects of free will belief manipulations on behaviour, we need to assess beliefs and attitudes within the same participants. This would also allow us to test whether free will attitudes are conditional on specific free will beliefs (e.g. ‘I can only value free will if I believe in it’), or vice versa. Relatedly, data reported here have been acquired between 2005 and 2007, while much of the work on free will beliefs has been performed about a decade later [4,5]. Attitudes can change across time, which further emphasizes the necessity to acquire beliefs and attitudes simultaneously if we are to understand how these two aspects of lay views on free will interact.
Posts
No comments:
Post a Comment