Elsevier

Consciousness and Cognition

Volume 25, April 2014, Pages 27-41
Consciousness and Cognition

The free will inventory: Measuring beliefs about agency and responsibility

https://doi.org/10.1016/j.concog.2014.01.006Get rights and content

Highlights

  • We summarize and criticize the existing tools for measuring free will beliefs.

  • We present a new tool we constructed for measuring free will beliefs.

  • We discuss potential uses for our new tool for measuring free will beliefs.

Abstract

In this paper, we present the results of the construction and validation of a new psychometric tool for measuring beliefs about free will and related concepts: The Free Will Inventory (FWI). In its final form, FWI is a 29-item instrument with two parts. Part 1 consists of three 5-item subscales designed to measure strength of belief in free will, determinism, and dualism. Part 2 consists of a series of fourteen statements designed to further explore the complex network of people’s associated beliefs and attitudes about free will, determinism, choice, the soul, predictability, responsibility, and punishment. Having presented the construction and validation of FWI, we discuss several ways that it could be used in future research, highlight some as yet unanswered questions that are ripe for interdisciplinary investigation, and encourage researchers to join us in our efforts to answer these questions.

Introduction

Philosophers have long debated the nature of free will and moral responsibility. Psychologists have more recently joined the fray. While some researchers have tried to shed empirical light on the debate concerning whether we have free will, typically concluding that we do not (e.g., Bargh, 2008, Harris, 2012, Libet, 1999, Wegner, 2002), others have focused instead on exploring our beliefs about free will and the role they play in our daily lives (e.g., Alquist et al., 2013, Bergner and Ramon, 2013, Haynes et al., 2003, Nettler, 1959, Nettler, 1961, Ogletree and Oberle, 2008, Pronin and Kugler, 2010, Schooler et al., in press, Stillman and Baumeister, 2010, Stillman et al., 2010, Stroessner and Green, 1990, Viney et al., 1988, Viney et al., 1982, Waldman et al., 1983). Each of these two projects is interesting and important in its own right. For present purposes, we focus on the psychology of believing in free will—a topic that has received increasing attention in the wake of gathering evidence that challenging people’s beliefs about free will may influence their behavior in surprising and sometimes alarming ways. For example, telling people they do not have free will has been shown to increase cheating (Vohs & Schooler, 2008) decrease helping behavior and increases aggression (Baumeister, Masicampo, & DeWall, 2009), reduce self-control (Rigoni, Kuhn, Gaudino, Sartori, & Brass, 2012), and impair detection of errors (Rigoni, Wilquin, Brass, & Burle, 2013). These findings suggest that believing in free will may be instrumentally valuable from the standpoints of positive psychology and public morality. This is true regardless of whether or not we actually have free will, a related but orthogonal issue that we will not be exploring here.

These recent findings also highlight the importance of having valid and reliable tools for measuring beliefs about free will and related concepts. For while the gathering data suggest that diminishing people’s belief in free will may lead to increased cheating, decreased helping, lower punishment judgments, and the like, the validity of these findings depends in part on the validity and reliability of the scales used to measure people’s beliefs about free will, determinism, and responsibility. As such, these otherwise exciting findings have highlighted a problem that has hindered empirical research on free will beliefs from the outset—namely, that while researchers have developed several scales during the past thirty years for measuring beliefs about free will (e.g., Paulhus and Carey, 2011, Paulhus and Margesson, 1994, Rakos et al., 2008, Stroessner and Green, 1990, Viney et al., 1984, Viney et al., 1982), each of these tools is problematic. The main goal of our project was to develop a new psychometric instrument that would have greater validity than previous scales while at the same time yielding more fine-grained data concerning how people think about the complex relationships among free will, responsibility, dualism, choice, determinism, and related concepts.

In this paper, we first discuss the extant tools that have been developed for measuring free will beliefs, and we identify some of their limitations. Then, we present the results of the construction and validation of a new psychometric tool for measuring beliefs about free will: The Free Will Inventory (FWI). The FWI is a 29-item instrument with two parts (which can be given together or separately). Part 1 consists of three five-item subscales designed to measure strength of belief in free will, determinism, and dualism. Part 2 consists of a series of 14 statements designed to further explore people’s associated beliefs and attitudes about free will, (in)determinism, choice, the soul, predictability, responsibility, and punishment. After presenting the construction and validation of FWI, we (a) discuss several ways that it could be used in future research, (b) highlight some as yet unanswered questions that are ripe for interdisciplinary investigation, and (c) encourage researchers to join us in our efforts to answer these questions.

Most of the earliest work on the psychology of believing in free will focused on the relationship between beliefs about free will, determinism, and punishment (e.g., Nettler, 1959, Nettler, 1961, Stroessner and Green, 1990, Viney et al., 1982, Viney et al., 1988). Though groundbreaking, the results were often mixed and hard to interpret. For instance, in one of the first studies on free will beliefs, Nettler (1959) claimed to have found that believing in free will is correlated with cruelty, retribution, and revenge and that believing in determinism is correlated with people being less punitive and treating others with more dignity. In response, Viney et al. (1982) first reported data that seemed to support the opposite conclusion—namely, that people who believe in free will are less rather than more punitive—before later finding “neither reliable correlations between punitiveness and beliefs in free will or determinism nor reliable correlations between rationales for punishment and beliefs in free will or determinism” (Viney et al., 1988, p. 20).

Looking back on the early empirical work on free will beliefs with the benefit of hindsight, there are at least two lessons to learn. First, what researchers find when it comes to beliefs about free will can depend a great deal on how the researchers conceptualize and think about free will. Not only can researchers’ own free will beliefs influence how they design their experiments, phrase their questions, etc., but these beliefs may also color how the findings are analyzed and interpreted. Let us call this the problem of theory contamination—i.e., when researchers’ own theoretical commitments unduly influence or bias their findings. While it may be impossible to keep one’s theoretical commitments entirely at bay while designing studies and analyzing data, it is important for researchers to make a concerted effort to avoid theory contamination as much as possible. One strategy that we adopted was making sure we had both psychologists and philosophers on our team, as well as people with competing views about free will and its relationship to determinism.

A second lesson to be learned from the early research on free will beliefs is that researchers should have a shared vocabulary (or lexicon) when exploring these types of complex beliefs and attitudes. Instead, much of the early empirical work in this area contains conflicting definitions of key terms (or usages of key terms) such as determinism, libertarian free will, choice, dualism, fatalism, and the like. Moreover, not only are key terms used in ways that differ from one researcher to the next, these terms are often defined in ways that most philosophers who specialize in the free will debate would reject. This generates worries about both content and criterion validity. While philosophical experts on free will should not have carte blanche to define contested terminology however they see fit, the long and deep engagement of philosophers with the question of free will should be taken into consideration during empirical investigations of free will beliefs.

At least some of these methodological worries might be alleviated if psychologists and philosophers worked together rather than laboring individually on their own respective sides of the disciplinary divide. For while psychologists do not appear to have a shared lexicon when it comes to free will and related concepts, philosophers have developed a common vocabulary for talking about these issues. Take, for instance, the concept of determinism. While the term has a variety of meanings and uses (both in everyday life and amongst psychologists), philosophers typically have one thing in mind when they discuss determinism—namely, the thesis that given the actual past and the laws of nature, there is only one possible future at any moment in time (e.g., van Inwagen, 1983).

Using this definition of determinism, the dominant issue in the philosophical debate about free will has traditionally been whether free will and moral responsibility are compatible with determinism. The attempt to address this so-called “compatibility question” has generated a wide variety of positions. Incompatibilists claim that free will and determinism cannot coexist. Among incompatibilists, libertarians argue that humans have free will, and hence deny the truth of determinism, some arguing that indeterminism helps secure free will (e.g., Kane, 1996), others arguing that free will requires a special causal power, agent-causation (e.g., O’Connor, 2000). Other incompatibilists argue that we lack free will, either because determinism is true (hard determinists) or, more commonly, because they argue that free will and moral responsibility are incompatible with indeterminism as well as determinism (hard incompatibilists or skeptics about free will; e.g., Pereboom, 2001, Strawson, 1994).

However, a long philosophical tradition has defended compatibilism about free will and determinism. Compatibilists offer various analyses of the capacities required to be free and responsible agents, such as our abilities to identify with some of our desires over others (Frankfurt, 1988), to act in accord with moral reasons (Wolf, 2000), to be appropriate targets of reactive attitudes such as indignation or approbation (Strawson, 1962), or to be appropriately responsive to reasons (Fischer & Ravizza, 1998). In general, compatibilists argue that free will and moral responsibility do not require the unconditional ability to do otherwise, holding fixed the actual past and laws, nor the seemingly impossible ability to be ultimately responsible for what makes us the way we are (see Strawson, 1994). Instead, compatibilists argue that free and responsible agency requires the capacities involved in self-reflection and practical deliberation; free will is the ability to make choices based on reasons, along with the opportunity to exercise this ability without undue constraints.

Despite the fact that compatibilism is the majority position among philosophers (Bourget & Chalmers, in press) and despite the shared vocabulary that philosophers use to discuss free will, psychologists have often operationalized free will and determinism in ways that are not only discordant with the philosophical tradition but which also preclude a combination of beliefs that is not only logically possible but also commonly held (at least by certain groups of people). This has led to some conflicting research programs and unclear results. To see what we have in mind, consider the earliest attempt to develop a scale to measure beliefs about free will, the Free Will and Determinism Scale (FWD) presented in Viney et al. (1982). The general instructions for FWD read as follows:

For centuries, human beings have debated the old philosophical problem of free will and determinism. Free will is the doctrine that assumes that we make real choices which are partially or completely independent of antecedent conditions. Determinism denies this and maintains that causation is operative in all human affairs. Thus, so-called choices are influenced or determined by antecedent conditions. The following materials include statements which represent opinions by those who believe in free will and those who believe in determinism. Please read each of the following statements carefully and then place an x under the statement which most closely corresponds to your opinion on the topic. If your opinion seems to fall somewhere between the statements provided, then mark an x in one of the intermediate spaces.

These instructions highlight the aforementioned worries about theory contamination.

After all, it is clear that Viney et al. (1982) have simply built the assumption of incompatibilism into FWD from the outset, in part by defining ‘determinism’ in contrast with ‘free will’. As such, there is no way to use FWD to tease apart individuals who have incompatibilist intuitions from those who have compatibilist intuitions. Indeed, at some point, Viney and colleagues invite participants to choose between the following two statements: (a) “I believe strongly in free will,” and (b) “I believe strongly in determinism.”

Given the way that FWD is constructed, individuals who do not take determinism to rule out free will have no way to express that belief. Furthermore, if someone expresses belief in free will but also expresses the belief that determinism is false (e.g., because of quantum theory), that does not indicate that she is an incompatibilist, since she might think the truth or falsity of determinism is not relevant to whether we have free will (as do most contemporary compatibilist philosophers). The worry is not only theoretical—there is accumulating evidence that compatibilist intuitions are more widespread than philosophers and psychologists have traditionally assumed (e.g., Deery et al., in preparation, Monroe and Malle, 2010, Murray and Nahmias, 2012, Nadelhoffer, in press, Nadelhoffer and Tocchetto, 2013, Nahmias et al., 2006, Nichols, 2004, Nichols, 2006, Nichols and Knobe, 2007, Shepard and Reuter, 2012, Shepherd, 2012; Rose & Nichols, 2013). Shedding light on the scope and limits of these compatibilist beliefs and attitudes is at least one of the issues researchers ought to examine. But if the tool we use for measuring people’s free will beliefs is predicated on incompatibilism, we cannot do so.

Unfortunately, the same criticism applies to a more recently developed and commonly used psychometric tool for measuring free will beliefs: the Free Will and Determinism Scale (Filevich et al., 2013, Rakos et al., 2008, Rigoni et al., 2011, Rigoni et al., 2012). In the development of their scale, Rakos and colleagues acknowledged that they followed in Viney’s incompatibilist footsteps by conceptualizing free will and determinism “as the opposing poles of an ‘amount of agency’ continuum” (p. 25). Moreover, there are additional worries about the scale designed by Rakos and colleagues. For instance, the validation round involved less than 100 participants, and thus was likely underpowered (Kline, 2011). Moreover, each of the items in their free will subscale contains the term “free will”—which may artificially inflate how well the free will items appear to hang together (Chen & Thissen, 1997). Finally, it is problematic that there is so much overlap between the items in the various subscales. For instance, why is “My decisions are influenced by a higher power” included within the Personal Limitations subscale rather than the Higher Power Control subscale? But for present purposes, the main shortcoming of the scale designed by Rakos and colleagues is the aforementioned worry that it, too, is predicated on incompatibilism.

Paulhus and Carey (2011) recently published a scale that we believe does a better job of avoiding the incompatibilist assumption while at the same time avoiding some of the other problems we have discussed—e.g., small samples, subscale items that all have the same word or term, etc. The Free Will and Determinism Scale (FAD-Plus) is a refinement of an earlier version of the scale (FAD, Paulhus & Margesson, 1994).1 FAD-Plus is a 27-item scale that consists of four primary subscales: (a) Free Will (e.g., People have complete free will); (b) Scientific Determinism (e.g., As with other animals, human behavior always follows the laws of nature); (c) Fatalistic Determinism (e.g., Whether people like it or not, mysterious forces seem to move their lives); and (d) Unpredictability (e.g., Chance events seem to be the major cause of human history).

Contrary to the incompatibilist assumption that influenced the research described above, Paulhus and Carey (2011) found that responses to free will items and responses to scientific determinism items reliably loaded onto largely independent factors. Moreover, over multiple rounds involving several hundred participants, Paulhus and Carey consistently found that aggregate scores on the free will subscale and the scientific determinism subscale tended to be neither positively nor negatively correlated.

In light of these findings, Paulhus and Carey perhaps unsurprisingly conclude that their work on FAD-Plus “supports the conclusion that lay judges see freewill and determinism as quite compatible” (2011, p. 102). However, we prefer to remain agnostic for present purposes. The existing findings on free will beliefs are still too messy and inconclusive to draw any firm conclusions when it comes to the debate about whether most people have incompatibilist beliefs about free will. We need look no further than the data from FAD-Plus for an illustration of just how murky things can be. For instance, scores on their Fatalistic Determinism subscale correlate with scores on the Unpredictability subscale, which we find difficult to explain. More surprisingly, their Fatalistic Determinism subscale does not inversely correlate with the Free Will subscale, even though the former includes items such as, “What will be will be—there’s not much you can do about it” and “No matter how hard you try, you can’t change your destiny.” It seems that people who agree with these items, which suggest significant lack of control, would more strongly disagree with items, such as “People have complete control over the decisions they make,” and vice versa. In any case, given the content of the items, it is unclear what exactly the Fatalism and Unpredictability subscales are measuring.

There are additional problems with FAD-Plus as well. For instance, three of the seven Free Will items are about responsibility and blame, such as “Criminals are totally responsible for the bad things they do.” That these items tend to statistically load with items describing control and free will is an interesting finding (and below we will describe similar patterns in our early rounds of data collection), but such items seem to be indirect measures of beliefs about free will and control, and we think it is more useful to have ways to examine the relationships between beliefs about free will and beliefs about responsibility, rather than building them into the same subscale. Furthermore, the Scientific Determinism subscale seems designed to describe the idea that people’s character and actions are caused by their genes and upbringing, rather than describing the philosophical concept of determinism (e.g., the idea that there are sufficient causes for every event, including human decisions). Discovering that beliefs about these types of causation do not correlate with beliefs about free will is important, but it may not inform us about whether people have incompatibilist beliefs. Indeed, even though FAD-Plus does not presuppose incompatibilism as earlier scales did, it does not provide any obvious way to distinguish people who are disposed towards incompatibilism from those who are more inclined toward compatibilism. One could look for people who have high scores on both the Free Will subscale and the Scientific Determinism subscales as a proxy for compatibilism. But this may underestimate compatibilist beliefs since it mistakenly assumes that compatibilists believe strongly in determinism—which certainly need not be the case; and they certainly do not have to agree with statements suggesting, for instance, that our actions are determined by our genes. An indeterminist can still be a compatibilist—which requires merely that one believes that if determinism were true, we could be free. Consequently, trying to distinguish compatibilists from incompatibilists by comparing their aggregate subscale scores from FAD-Plus is problematic. Third, FAD-Plus fails to provide us with a way of shedding any light on the relationship between people’s free will beliefs and their beliefs about the soul, the mind, and the brain more generally. Finally, FAD-Plus also does not enable us to explore people’s beliefs about free will, determinism, and dualism and their beliefs about punishment and responsibility.

While we think FAD-Plus is better, all things considered, than the earlier incompatibilist scales, the need remains for a scale that describes the relevant concepts in a more precise and fine-grained way, ideally lining up more closely with the philosophical tradition, and that examines relationships between a wider range of relevant beliefs. At a minimum, none of the existing scales can shed much light on the aforementioned debate about whether or not most ordinary people have beliefs about free will that support incompatibilism.2 In light of the flaws and limitations of the existing psychometric tools, we developed a new instrument for measuring beliefs about free will and related concepts.

Section snippets

Developing and Refining Part 13 of FWI4

Our goal was to adopt an approach to scale construction that was as theory-neutral as possible in the hopes that it would help us minimize the pitfalls of theory contamination. For instance, rather than present participants with a forced incompatibilist choice between free will and determinism, we wanted to present them instead with a variety of statements that would capture a wider range of beliefs and attitudes. Another important decision we made was that we would recruit people from the

Validating FWI with confirmatory factor analysis

Many lessons were learned from the four development rounds. For example, it is not clear if determinism and fatalism, or if free will and moral responsibility, form clear and distinct conceptual categories for most people. It is possible that most people do distinguish determinism from fatalism, and free will from moral responsibility, but given the close relationships between the concepts within these pairs, it may be difficult for psychometric tools to tease apart these concepts (one goal of

Developing and refining Part 2 of FWI

One of our primary goals was to design a psychometric instrument for measuring the strength of people’s beliefs in free will, determinism, and dualism. In doing so, we also wanted to avoid some of the limitations and problems of the extant tools. After five rounds of data collection and analysis, we believe we were largely successful in accomplishing these goals. Not only were we able to get items that measured the factors we were interested in exploring, but we were able to do so in a way that

General discussion

Increasingly, scientists argue that their discoveries challenge the existence of free will. Meanwhile, as we saw earlier, several studies indicate that informing people about such discoveries can change the degree to which they believe in free will and subtly alter some of their behaviors. But the psychometric tools used in these studies to measure people’s beliefs about free will are arguably problematic (see Section 1). So, we decided a new instrument was needed (especially given the

Acknowledgments

This work was supported by the John D. and Catherine T. MacArthur Foundation, The Law and Neuroscience Project, and The Regents of the University of California (TN), a Big Questions in Free Will (BQFW) grant from the John Templeton Foundation (TN and EN), and a Neuroethics Scholar Program Fellowship from Emory University’s Center for Ethics (JS). The opinions expressed in this publication are those of the authors alone and do not necessarily reflect the views of the John D. and Catherine T. MacArthur

References (68)

  • P.M. Bentler

    EQS structural equations program manual

    (1993)
  • R.M. Bergner et al.

    Some implications of beliefs in altruism, free will, and nonreductionism

    The Journal of Social Psychology

    (2013)
  • Bourget, D., & Chalmers, D. (in press). What do philosophers believe? Philosophical...
  • M.W. Browne et al.

    Alternative ways of assessing model fit

  • B.M. Byrne

    Structural equation modeling with EQS and EQS/Windows: Basic concepts, applications and programming

    (1994)
  • R.B. Cattell

    The scree test for the number of factors

    Multivariate Behavioral Research

    (1966)
  • R.B. Cattell et al.

    A general plasmode (No. 30-10-5-2) for factor analytic excercises and research

    Multivariate Behavioral Research Monographs

    (1967)
  • W. Chen et al.

    Local dependence indexes for item pairs using item response theory

    Journal of Educational and Behavioral Statistics

    (1997)
  • D.A. Cole

    Utility of confirmatory factor analysis in test validation research

    Journal of Consulting and Clinical Psychology

    (1987)
  • Deery, O., Bedke, M., & Nichols, S. (in preparation). Phenomenal abilities: Incompatibilism and the experience of...
  • L. Fabrigar et al.

    Evaluating the use of exploratory factor analysis in psychological research

    Psychological Methods

    (1999)
  • J.L. Fava et al.

    The effects if over extraction on factor and component analysis

    Multivariate Behavioral Research

    (1992)
  • J.M. Fischer et al.

    Responsibility and control: A theory of moral responsibility

    (1998)
  • H. Frankfurt

    The importance of what we care about

    (1988)
  • J.F. Hair et al.

    Multivariate data analysis with readings

    (1998)
  • S. Harris

    Free will

    (2012)
  • S.D. Haynes et al.

    Free will, determinism, and punishment

    Psychological Reports

    (2003)
  • C.J. Hopwood et al.

    How should internal structure of personality inventories be evaluated?

    Personality and Social Psychology Review

    (2010)
  • H.F. Kaiser

    The application of electronic computers to factor analysis

    Educational and Psychological Measurement

    (1960)
  • R. Kane

    The significance of free will

    (1996)
  • R.B. Kline

    Principles and practice of structural equation modeling

    (2011)
  • B. Libet

    Do we have free will?

  • R.C. MacCallum et al.

    Power analysis and determination of sample size for covariance structure modeling

    Psychological Methods

    (1996)
  • A.E. Monroe et al.

    From uncaused will to conscious choice: The need to study, not speculate about people’s folk concept of free will

    Review of Philosophy and Psychology

    (2010)
  • Cited by (121)

    • The impact of free will beliefs on implicit learning

      2023, Consciousness and Cognition
      Citation Excerpt :

      Three separate exclusion criteria were then applied. First, we tested for missing data in the free will inventory (FWI, (Nadelhoffer et al., 2014)). Secondly, 3 independent raters screened the responses in the free will manipulation procedure for low-effort responses.

    View all citing articles on Scopus
    View full text