Our overall project: how to solve the psychological problem of theological fatalism?

Folk theories are made of concepts and (implicit) principles that allow people to make sense of their environment (Gelman and Legare Reference Gelman and Legare2011). Folk biology, for instance, includes concepts such as the concepts of birth, growth, inheritance, innate nature (Griffiths et al. Reference Griffiths, Machery and Linquist2009), and disease (Machery Reference Machery, Hens and de Block2023), as well as principles that allow people to make sense of the biological world, for example the existence of species and the similarity between parents and offspring. Theory of mind allows people, among other things, to explain and predict behaviour by deploying concepts such as the concepts of belief and desire (e.g. Nichols and Stich Reference Nichols and Stich2003), and principles, such as those connecting perceptual beliefs and lines of sight (e.g. Povinelli and Eddy Reference Povinelli and Eddy1996).

In addition to folk theories such as folk biology, folk physics, and theory of mind, people seem to have a folk epistemology, viz. a theory of what it means to know something or have justified beliefs (e.g. Gerken Reference Gerken2017; Heintz and Taraborelli Reference Heintz and Taraborelli2010; Kitchener Reference Kitchener2002; Machery et al. Reference Machery2017). This folk epistemology includes the concept of knowledge, and some epistemic principles (e.g. that one cannot know by sheer luck) appear to be surprisingly robust across cultures (Machery et al. Reference Machery2017). Text-analytic evidence also suggests that the concept of knowledge might be infallibilist, at least for English speakers: Nichols and Pinillos (Reference Nichols and Pinillos2018) report that in the CHILDES corpus occurrences of ‘to know that’ are never accompanied by an expression of fallibility such as ‘I am not sure’, in contrast to occurrences of ‘to think that’. That is, in child-directed speech, parents appear to be uttering sentences such as (1), but not sentences such as (2):

1. 1) I think the bread is on the table. I am not sure.

2. 2) I know the bread is on the table. I am not sure.

By appealing to Bayesian ideas about confirmation, Nichols and Pinillos conclude that children are likely to acquire an infallibilist concept of knowledge: Since no expression of fallibility comes with utterances of ‘to know that’, the likelihood that this expression expresses an infallibilist concept of knowledge is higher than the likelihood that it expresses a fallibilist concept of knowledge.

Psychologists have also examined extensively, often in a developmental context, how people detect and understand agency (e.g. Johnson et al. Reference Johnson, Slaughter and Carey1998). When we view others as agents, we view them as having goals and as being in control of their actions, which fulfil their goals in an efficient and rational manner (Csibra et al. Reference Csibra, Gergely, Bíró, Koos and Brockbank1999; Woodward Reference Woodward1998). Recently, Machery and colleagues (Machery Reference Machery, Hens and de Block2023) have shown that merely showing visual cues of agency such as continuous, self-initiated motion is enough to lead people to assign free will and self-control to a marble, exactly as visual cues lead people to assign beliefs and desires to a marble (Heider and Simmel Reference Heider and Simmel1944).

Sometimes the domains of folk theories overlap, and the same events and processes can be conceptualized in different ways. Thus, folk epistemology and the folk theory of agency both apply to human actions. We can reason about the epistemic status of beliefs about past, present, and future actions; we can assign knowledge of past, present, and future actions: You probably know that G.W. Bush decided to invade Iraq; Edouard Machery assigns to his coauthor, Ameni Mehrez, the knowledge that he is right now editing this very sentence; finally, Ameni Mehrez assigns to Edouard Machery the knowledge that many French families will go to the beach during August.

The conceptualizations rooted in folk theories can inhibit one another, and taking one perspective about an event or a process (applying the concepts and principles constitutive of one theory) can prevent one from taking another perspective. For instance, viewing an action as a physical process seems to stand in the way of viewing it as a genuine action. Jack and colleagues (Jack et al. Reference Jack2013) examined the neural correlates of mechanical thinking and of social cognition, and found mutual inhibitions: Solving physical puzzles inhibited the neural correlates of social cognition and vice-versa.

We hypothesize that folk epistemology and the folk theory of agency cannot be easily brought to bear on a future action simultaneously: If one views an event as a human action, one views it as free and one thinks of the agent as responsible for this action (Machery et al. Reference Machery, Kneer, Willemsen, Newen, Murray and Henne2023); if for the folk, freedom requires the capacity to have done otherwise, the agent could thus have done otherwise, and the action didn’t have to occur; but if one views it as known in advance, one is inclined to view it as unavoidable, and thus as not free. Of course, from a philosophical point of view, foreknowledge is arguably consistent with free will if knowledge isn’t infallible:Footnote ² If so, knowledge that an agent will act in a particular way is consistent with the fact that this agent could act differently (although it is not consistent with them actually acting differently). But first, as noted earlier, the folk concept of knowledge might well be infallibilist, and if it is, foreknowledge, as conceived by lay people, might be incompatible with free will; second, even if the folk concept of knowledge isn’t infallibilist, it might still be psychologically difficult to view a future action as both known and free. To distinguish it from the philosophical problem, we call this difficulty the ‘psychological problem of theological fatalism’.

This hypothesized tension between a folk-epistemological and a folk-agentic perspective on human action might be the source of the philosophical problem of theological fatalism. (In fact, many philosophical problems might be rooted in tensions in lay cognition; Porter et al. Reference Porterin press; Rose et al. Reference Rose2020.) Philosophers and theologians carefully elaborate the commonsensical puzzlement about how an action can be free but infallibly known in advance, as it must be if God has infallible foreknowledge of future actions.

In contrast to philosophers’ sophisticated discussion of the philosophical problem, lay believers can alleviate the tension between divine foreknowledge and free will in lay cognition by denying free will or by curtailing divine foreknowledge. Our overall research about religious traditions (the first part of which is reported in this article) is rooted in the hypothesis that different religious traditions prefer one of these two options. Specifically, we hypothesize that lay believers within Sunni IslamFootnote ³ and Christian denominations influenced by Calvinism deny, or at least deemphasize, free will and assert divine foreknowledge, while other traditions such as Catholicism emphasize free will, and find ways to curtail, or at least deemphasize, divine foreknowledge.Footnote ⁴ Lay believers, if not philosophers and theologians, might thus alleviate the psychological problem of theological fatalism differently.

Available evidence relevant to the hypothesis guiding our overall research is mixed. Consistent with the hypothesis that Sunni Islam deemphasizes free will, among Turks, religiosity correlates with the subscale of the Free Will and Determinism Scale (‘FAD-plus’) called ‘fatalistic determinism’, which uses items such as ‘I believe that the future has already been determined by fate’ and ‘Fate already has a plan for everyone’ (Yilmaz et al. Reference Yilmaz, Bahçekapili and Harma2018). A third of a sample of 347 Muslim Americans agreed that ‘Everything in life is determined by God’ (Haddad and Lummis Reference Haddad and Lummis1987). According to a PEW study conducted between 2011 and 2012 (Lugo et al. Reference Lugo, Cooperman, Bell, O’Connell and Stencel2012), in nineteen out of the twenty-three countries surveyed, seven out of ten Muslims said they believe in fate. The numbers are the highest for countries like Tunisia (98 per cent), Egypt (93 per cent), and Jordan (91 per cent), and the lowest in Albania (44 per cent) and Kosovo (50 per cent). Muslims are also more fatalist than Christians with respect to poor health (e.g. Baron-Epel et al. Reference Baron-Epel, Friedman and Lernau2009; Hess and McKinney Reference Hess and McKinney2007; Pipes Reference Pipes2015). According to the World Value Survey conducted in 2004 (Acevedo Reference Acevedo2008), Muslims are more fatalistic than Catholics and Protestants.Footnote ⁵

Turning to Christians, consistent with the hypothesis that Calvinism might deemphasize free will, Weber (Reference Weber1930) famously highlighted the significance of Calvinists’ views about predetermination in their economic success and in their contribution to the birth of capitalism. Van Elk and colleagues (Van Elk et al. Reference Van Elk, Rutjens and van Harreveld2017) found that protestants have weaker beliefs in free will than Catholics. On the other hand, using data from the World Value Survey, Ruiu (Reference Ruiu2013) found little difference in the influence of religious affiliation on fatalism,Footnote ⁶ although religiosity in general promoted fatalism.

The present study: folk beliefs about divine foreknowledge across religions

It may seem obvious, particularly to believers, that the faithful endorse divine foreknowledge, but do they really, and do they equally across Abrahamic religions and religious traditions?

Psychologists and anthropologists have proposed that a belief in ‘big Gods’ (aka ‘moralizing Gods’) has played an important role in the evolution of cooperation in large groups, a characteristic of human hypersociality (Norenzayan Reference Norenzayan2013; Norenzayan et al. Reference Norenzayan2016). In a large group, it is difficult to distinguish people who are pulling their weight for the common good from free riders, and the former are thus at a disadvantage: In contrast to them, free riders reap the benefit of others’ cooperation without paying any cost. Cooperation is thus threatened. A belief in knowing, powerful, and just Gods (‘big Gods’) is hypothesized to motivate potential cooperation among group members that would otherwise be free riders: Since Gods know how group members behave and can punish them for their actions, potential free riders are incentivized to cooperate (for relevant evidence, see Lang et al. Reference Lang2019). The big-Gods hypothesis about the evolution of human cooperation assumes that believers assign extensive knowledge of human actions to Gods; in this respect, the Abrahamic God, omniscient and omnipotent as it is, is the epitome of big Gods. While the big-Gods hypothesis emphasizes omniscience, it has in fact little to say about whether believers in big Gods, including the Christian and Muslim God, endorse divine foreknowledge. (In fact, ‘foreknowledge’ does not appear in Norenzayan’s influential book, Big Gods: how religion transformed cooperation and conflict.) Rather, the focus is on God’s knowledge of current actions (God is monitoring the decisions group members are taking) and past actions (God knows that group members have done and will reward or punish them accordingly).

Furthermore, it might be attractive to curtail God’s omniscience in some way in order to preserve free will. Indeed, in philosophy, a common, but controversial response to the challenge divine foreknowledge raises for free will is to deny that there are truths about future free actions and thus that God has foreknowledge of those. This might be the case because there are no truths about future contingents in general (Prior Reference Prior1962) or just about future free actions (Swinburne Reference Swinburne2017). Open theists, among others, hold that the future is undetermined because of human free will, and that God knows that it is undetermined (e.g. Pinnock et al. Reference Pinnock, Rice, Sanders, Hasker and Basinger1994). Curtailing foreknowledge partly or fully preserves God’s omniscience since God knows everything there is to know, but omniscience is only preserved because there is less to know than one might have thought.

So, it is not entirely clear whether we should expect lay believers to endorse divine foreknowledge. In the study reported in this article, we hypothesized that the salience of divine foreknowledge varies across religions or religious traditions. While all the religious traditions we consider accept both foreknowledge and free will, they disagree about how to understand their relation, and they emphasize foreknowledge to a greater or smaller extent. Sunni Islam often emphasizes God’s foreknowledge, as we noted earlier (Wan Zakaria Reference Wan Zakaria2015).Footnote ⁷ We expected Calvinism to be similar to Sunni Islam in this respect because of the connection between providence and predetermination: God predetermines and as a result foresees who will be saved and who will be damned. As Calvin put it (Institutes of the Christian Religion, 1.16.2), ‘nothing happens but what [God] has knowingly and willingly decreed’.Footnote ⁸ By contrast, Catholicism deemphasizes foreknowledge, while emphasizing free will and the responsibility of individuals. In the Catechism of the Catholic Church, ‘foreknowledge’ appears only once (II, 599, 155), and its significance is presented as follows:

Providence is extensively discussed, but as is the case of foreknowledge, discussions of providence highlight its compatibility with free will (II, 306, 81): ‘God is the sovereign master of his plan. But to carry it out he also makes use of his creatures’ cooperation. … God grants his creatures not only their existence, but also the dignity of acting on their own.’Footnote ¹⁰

Further, the Catechism emphasizes the significance of free will throughout (with more than 100 occurrences of ‘freedom’; see, e.g. Article 3 ‘Man’s freedom’, 430).

Finally, Methodists present a difficult case. Providence plays an important role for them since on their view God foresees but does not predetermine which individuals will be saved, freely accepting divine grace, and which individuals will be damned, freely rejecting divine grace.Footnote ¹¹ So, divine foreknowledge plays an important role in Methodist theology, but is always coupled to an emphasis on human free will and responsibility.

Finally, we hypothesized that agreement with divine foreknowledge is partly influenced by motivational factors. Assigning to an omnipotent being the foreknowledge that an agent will do something wrong might feel wrong to lay people: If God knows in advance that an agent will sin, why doesn’t God prevent this sin to happen? The philosophical problem of evil develops this commonsensical idea (Leibniz 1709/Reference Leibniz and Huggard1985; Mackie Reference Mackie1955). Perhaps, to avoid this question, lay believers might be less inclined to agree with divine foreknowledge of morally wrong actions.

In light of the considerations discussed in the previous paragraphs we preregistered the following four hypotheses:

Participants

The study was preregistered on OSF (doi.org/10.17605/OSF.IO/SKJA9), and we report below all the deviations from our preregistration. To determine sample size, we collected pilot data from 152 American participants recruited on Prolific and 43 Muslim participants sampled by snowball sampling on social media (Facebook). In addition to commenting on a pilot version of the final survey (described in the next section), their answers were used to estimate the relevant parameters of our model (for two religious groups only). On the basis of these estimates, we simulated increasingly large datasets (from 200 to 500 participants) using the simr package in R. Setting alpha at .005 as preregistered (Benjamin et al. Reference Benjamin2018), we determined that a sample size of 250 per religious group was sufficient to get a power larger than .8 to identify effects comparable to those estimated based on our pilot data. The pilot data, R code, and power curves are available at OSF (osf.io/ecdrj/).

We collected participants from five different religious groups: Muslims in Tunisia, Muslims in the USA, Calvinists in the USA, Catholics in the USA, Methodists in the USA.Footnote ¹² We examined Muslims because of the possibility that divine foreknowledge might be more compelling to them, compared to some Christian traditions. We sampled Muslims in two different countries and cultural contexts for two reasons: (1) If we found differences between Tunisian and American Christians, it wouldn’t be clear that the differences are due to religion rather than differences in their respective cultural contexts; (2) if we failed to find any difference between American Muslims and Christians, it might be because of their shared cultural context and in spite of the religious differences. We examined Calvinists because of their commitment to the doctrine of providence and predetermination and Catholics because of the qualifications attached to the divine providence in the Catholic Catechism, as explained earlier. Finally, we surveyed Methodists because while they emphasize foreknowledge, it is always in the context of defending human free will.

The Tunisian sample was collected in Tunis, Tunisia by the Tunisian polling company One-to-One for Research and Polling. Participants were surveyed at their home, and they were not paid. The four other samples were collected online by CloudResearch and Qualtrics Panels, with the exception of a subset of Calvinist participants. Participants were all born in the USA and all reported being 18 or above. Participants recruited by CloudResearch and Qualtrics Panels were paid a small amount for participation. CloudResearch and Qualtrics Panels were asked to balance gender and age to the extent it was possible. Because online data companies were unable to recruit a large enough sample of Calvinists, we ultimately decided to rely on snowball sampling to increase our sample size. In line with our power analysis, we aimed to collect 300 participants meeting our preregistered exclusion criteria (more on these below), but we were not always able to do this. We aggregate the data collected in the USA (by the two data companies and by snowball sampling).

Table 1 reports the characteristics of the five samples.

Table 1. Characteristics of the five samples (education: 1 = elementary school, 2 = high school, 3 = some college, 4 = college degree, 5 = MA, 6 = phd)

Our samples were very religious.Footnote ¹³ On a scale ranging from 1 (‘Definitely not true’) to 5 (‘Definitely true’), the average answer for each group was superior to 4 for the measure of God’s significance in participants’ personal life (‘intrinsic religiosity’), which averages three questions (e.g. ‘My religious beliefs are what really lie behind my whole approach to life’). The Tunisian participants show the lowest average in terms of organizational activities. This is to be expected since half of the sample is composed of women and in Islam women are less likely to engage in organizational practices such as going to the Mosque (Gilliat-Ray Reference Gilliat-Ray2010). Consistent with our data, existing studies show that differences exist even among Muslims, with American Muslim women being more engaged and more likely to attend the Mosque than Muslim women living in Muslim-majority countries (Jamal Reference Jamal2005).Footnote ¹⁴ Our American samples were also on average college educated, but our Tunisian sample was on average less educated.

Lay conceptions of divine foreknowledge across religions

We have provided evidence that at least very religious Catholics, Reformists, Calvinists, and Muslims are confident that God has foreknowledge of future human actions. As noted in the introduction and in Section 1, viewing God as having foreknowledge of human actions raises the psychological problem of theological fatalism; how lay believers confronts this problem will be a focus of future work. The hypothesis guiding our research is that from a psychological point of view it is difficult to think of the future as both known and open: If God knows the future, then it is not open, and human actions are not free. Whether these two perspectives about future action really inhibit one another will be examined in future research.

We found some differences in the acceptance of divine foreknowledge between religious traditions: American and Tunisian Muslims drew a sharper contrast between human and divine foreknowledge than American Catholics, Reformists, and Calvinists, while Reformists drew a slightly weaker contrast than Catholics. Tunisian Muslims agreed more with divine foreknowledge than any of the other groups. We need to be careful in interpreting these findings since the difference between the Tunisian sample and the American samples could be due to different response styles to surveys (with a greater tendency to use the extreme points of a scale, e.g. Hui and Triandis Reference Hui and Triandis1989). On the other hand, the fact that a similar effect was found among American Muslims and Tunisian Muslims (although to a much smaller degree in the former sample), and the fact that this effect was expected in light of the salience of divine omniscience in general and of divine foreknowledge in Islam suggest that the effect might well be genuine: Foreknowledge might be more salient among Muslims than among Christians. Future work will aim at assessing this result further; in particular, using text-analytic tools, we are currently examining whether themes related to foreknowledge are more common in Islamic religious texts than in Christian ones.

To our surprise, we failed to find any difference between Catholics and Calvinists. Based on Calvinists’ emphasis on foreknowledge and predetermination, we had expected Calvinists to be more likely than Catholics and Reformists to agree with divine foreknowledge. This expectation was not supported by the present study. Future work should aim at assessing further this finding (one of the goals of our ongoing text-analytic work). The greater salience of foreknowledge for Muslims (if our finding is confirmed) raises various questions: Is the psychological problem of theological fatalism more salient to Muslims than to other believers? Supposing that the problem of theological fatalism must somehow be solved, are Muslims more likely to jettison or deemphasize free will than other believers? Future work will examine such questions.

We also examined whether the valence of a future action influences the assignment of foreknowledge because we wondered whether people might be reluctant to assign foreknowledge of bad actions to God: If God foresaw someone’s sins, why didn’t God prevent it? This line of thought might have led people, we speculated, to assign less foreknowledge of bad actions. The moral nature of a future action did influence the assignment of human foreknowledge, with people being more likely to agree with human foreknowledge of morally good actions. While we had not predicted this effect (indeed, we expected people to agree more with human foreknowledge of neutral actions), with the benefit of hindsight, it can perhaps be explained. In contrast to neutral actions such as cleaning the house and going to the beach (two of the twelve examples used in our study), good actions are often viewed as the expression of the agent’s character, a manifestation of the fundamental attribution error (Doris Reference Doris2002; Harman Reference Harman1999), and it is natural to assume that a neighbour would then be in a better position to anticipate them compared to neutral actions. But, if this explanation is correct, why aren’t morally bad actions treated as morally good actions? They too, one might think, might be the expression of people’s character. Perhaps, people are less willing to assign foreknowledge of morally bad actions because these are rarely done openly: Few people sin in public.

In any case, whatever the explanation is, human and divine knowledge differed. The moral nature of a future action had no detectable effect on divine knowledge. In fact, contrary to our expectations, the three types of actions were treated rather similarly (the exception being Reformists, who treated neutral actions somewhat differently from Catholics), and there was little indication that the wrongness of an action led very religious individuals to refrain from assigning foreknowledge to God.

One might speculate that religious believers have already explicitly considered the problem of evil, and that they have already decided that God’s foreknowledge of evil actions is consistent with its perfection. More plausibly perhaps, religious people similar to the ones we sampled might just be committed to divine foreknowledge in general, and they need not decide for each action whether God foreknows it, taking into consideration its particular characteristics. In fact, we suspect that at least for some participants it might be wrong to treat a question about divine foreknowledge as an open question, one that might call for a negative answer: A true believer does not deliberate about whether God foreknows a future action.

Psychologists working on religious cognition have often noted that believers use templates meant for human agents to characterize God’s agency. For instance, Gods are intuitively represented as doing actions in a sequential manner or as moving from one place to another, as a human agent would do (Barrett and Keil Reference Barrett and Keil2016; Barrett Reference Barrett1998; for review, see, Reference Barrett2000). Representations of supernatural agents are often thought to be ‘minimally counterintuitive concepts’ (Barrett Reference Barrett2007; Boyer Reference Boyer2007): Believers tend to think of God or Gods as violating only a few properties represented by their intuitive concepts (e.g. of agents). By contrast, here we see that divine and human foreknowledge are treated differently. This result is consistent with finding that from an early age on, beliefs are assigned differently to humans and to God (Barrett et al. Reference Barrett, Richert and Driesenga2001; Knight et al. Reference Knight, Sousa, Barrett and Atran2004; Richert and Barrett Reference Richert and Barrett2005). How closely representations of supernatural agents follow representations of human agents might depend on the task: Some might elicit more reflective representations, others more intuitive. Or it might vary across domains: Representations of supernatural agents’ locations and motions might be more tightly constrained than representations of their cognitive and epistemic capacities.

Our empirical work is limited in important respects. We first note two methodological problems. First, sampling our target populations was much more difficult than expected, and we were unable to reach our target sample size of 300 Calvinists; we also had to rely on snowball sampling. Experimental philosophy of religion should consider strategies to sample from religious populations, if it is not to limit itself to sampling mainstream religious groups and to rely on samples composed of individuals with low religiosity. Second, we had very strict exclusion criteria (answering twelve comprehension questions), and it turned out to be difficult, both in the online study and in-person study, for participants to get all these questions right. The questions themselves were extremely easy, but participants might not have been highly motivated to be accurate. Turning to more a substantive limitation, we note that our conclusions only generalize to religious individuals. We do not know whether less religious Catholics or Muslims would have also wholeheartedly agreed with divine foreknowledge or would agree equally with divine foreknowledge of good and bad actions.

From a philosophical point of view, the main lesson of this work is that the philosophical problem of theological fatalism echoes how lay believers think about God in Abrahamic religious traditions. This philosophical problem is not an abstruse concern, only of relevance to theologians and philosophers of religions, but a personal issue that believers in the Abrahamic God must address.