Event-causal Indeterminism & Free Will (Ekstrom) Blog post featured image

The debate around free will and determinism remains a central issue in philosophy. Navigating this complex discourse, Laura W. Ekstrom, a philosophy professor at William and Mary, contributes a paper that addresses persistent criticisms of libertarian free will. In defending a new event-causal indeterminism of free will, Ekstrom’s paper focuses on the “disappearing agent” problem and the role of luck or randomness in acts of free will.

This paper is a serious attempt to reconcile the demands of agent control with the realities of indeterministic processes. As I examine the core of Ekstrom’s arguments and the structure of her paper, we may gain insight into her proposed solution to the core issues of libertarian free will and provoke thoughts concerning free will and indeterminism.


Paper and the Author:

Laura W. Ekstrom is a philosopher known for her work in the philosophy of religion and metaphysics, particularly on topics related to free will, personal identity, and the nature of persons.

Laura W. Ekstrom is a Philosophy Professor at William and Mary (as of November 2023)

Here is a link to the article

Paper Relevance:

Accounts of free will and moral responsibility are essential.

One of the key merits of Esktrom’s work is that it directly engages with common criticisms of libertarian free will, such as the “disappearing agent” problem and the issues of luck or randomness. By doing so, Esktrom attempts to create a better theoretical foundation for libertarianism.

Libertarian views of free will are often criticized for being overly mysterious (in the case of agent-causal accounts) or failing to give the agent enough control. Esktrom’s paper aims to navigate between these extremes, arguing for a version of event-causal libertarianism that can account for meaningful agent control without resorting to obscure or problematic metaphysical claims.

The Aim of the Paper:

Esktrom attempts to provide and defend a new event-causal indeterministic account of free will, claiming that her account is better than existing libertarian accounts in the literature, especially better than agent-causal libertarianism.

Basic Landscape of Free Will Positions

Concerning free will, one can either hold that causal determinism is true or false. If false, one believes in causal indeterminism. Determinism on free will states that since causation, mental or physical, is determined by prior causes, one’s will is also determined by a prior chain of causes.

Each position can be divided into two positions. Determinism and indeterminism can be either compatible or incompatible with free will. Determinism may threaten free will because one’s will may be determined by external factors. Similarly, indeterminism may threaten free will because one’s will may be a matter of chance. In both cases, the agent is not in control of her will.
Determinism is true:

  • Hard determinism – Determinism and free will are incompatible.
  • Soft determinism (Compatibilism)- Determinism and free will are compatible.

Indeterminism is true:

  • Hard Indeterminism – Indeterminism and free will are incompatible.
  • Soft Indeterminism (Libertarianism) – Indeterminism and free will are compatible.

Here, Esktrom is defending the soft indeterminism (Libertarianism).

What is Indeterministic Causation According to Esktrom?

The author cautions us that free will does not necessarily follow from non-deterministic causation. Even if indeterminate, our will could be guided by external factors or be subject to chance.

The essential point about indeterministic causation (given that it exists) is that certain events can have a causal relationship with their outcomes without making them inevitable.

Esktrom expounds on two main approaches to indeterministic causal connections.

The first approach posits that causation involves probability-raising, meaning that an event (C) causes another event (E) if the occurrence of (C) makes (E) more likely to happen. This approach aligns well with an indeterministic account of free will, as the probabilities involved may be less than 1, allowing for (E) to not occur in the presence of (C).

Formally, here is the first approach:

is a cause of E in case P(E | C) > P(E | ∼C).

There are problems in this approach, but various modifications have been proposed that aren’t relevant here. However, the general characterization remains the same.

The second approach looks at causation in terms of physical processes or production, arguing that to cause is to produce, independent of probabilities. A chemical reaction is an excellent example of process causation.

Some theorists opt for hybrid views that integrate both probability-raising and production processes. Esktrom’s approach in this paper seems hybrid, but there is undoubtedly a probability-raising component. She explains her old view on event-causal indeterminism account of free will in section two (Old view); then, she shows the inadequacy of the view and, in section three, provides a revised account.

See here for more details about the event-causal account of free will

What is Esktrom’s Account of Free Will?
Event-causal indeterminism

In section two, Esktrom explains her old view on the event-causal indeterminism account of free will. She demonstrates the inadequacy of the view and, in section three, provides a revised account. Let us only concern the revised version here.

The author employs a modified version of Harry Frankfurt’s notion of the self to posit our ability to desire and shape preferences. Frankfurt notes that individuals possess the capacity to evaluate and shape their own desires by forming second-order desires—desires about their desires. He suggests that a person’s true self is defined by the second-order desires that inform their actions. This desires of desire is important in explaining how one can non-deterministically shape one’s preferences, beliefs, values, and desires.

For Esktrom, free actions encompass four main aspects: self-direction, non-deviant causation, absence of compulsion, and alternative possibilities. The author arrives at a theory of free action where one acts freely if one acts from a decisively formed preference or agent’s reasons (including beliefs, values, and desires) that could have been otherwise. This preference must be non-compulsively, non-deviantly developed and maintained. Moreover, the author explains that preference formation has an indeterministic causal history.

Below is Ekstrom’s specification of self-direction, non-deviant causation, absence of compulsion, and alternative possibilities:

  • Self-Direction: Esktrom develops an idea of the self based on “preferences” and “acceptances,” which are desires and beliefs formed with a specific aim—to desire what is good and to assent to what is true, respectively. The author refers to these as “decisively adopted or maintained,” implying a form of active agency rather than a passive one. These attitudes are essential in defining a core self or character that influences one’s actions.
  • Alternative Possibilities: Regarding alternative possibilities, the author discusses the importance of having the capacity to do otherwise for an act to be considered free. The author argues that the indeterministic formation of preferences allows the self to be different, fulfilling the requirement of alternative possibilities. Indeterminism provides the ability to do otherwise. This enhances control because the ability to do otherwise increases control.
  • Non-deviant causation: The action is caused non-deviantly and indeterministically by certain agent-involving events. This means that the action arises from the agent’s intrinsic properties rather than being caused by some external or unrelated factors. Even without a conscious decision, a “standing preference” might have been formed earlier and now guides the action. In other words, subconscious decisions flowing from preference count as free, according to Esktrom.
  • Absence of Compulsion: The action does not result from any form of compulsion, manipulation, or coercion that the agent hasn’t arranged for herself. This ensures that external forces do not compromise the agent’s will.

My interpretation of her view is that indeterminate causation is only required during preference formation. She and others acknowledge that if indeterminism plays a role in the direct causation of a decision or action, it will reduce the agent’s control (Laura Ekstrom (2000: ch. 4 and 2003) and Alfred Mele (1995: ch. 12, 1996, 1999b, and 2006: 9– 14)). Therefore, actions at any instance are determined by the causal chain. However, she argues that indeterminism occurring at earlier phases in the decision-making process may not necessarily have this effect.

Therefore, a person cannot alter indeterministic causal history at a whim but must build it. In other words, one’s act is free indirectly through indeterministic formed preferences, non-deterministically caused, and deterministically caused a decision and subsequent action. Moreover, preferences are central in shaping the probabilities associated with different actions. Preferences and other factors like beliefs and external circumstances influence the likelihood of choosing one step over another. However, they do not make the choice inevitable. The interpretation that every decision is indeterministic and free independently of background preferences is inconceivable and not well defended if this is correct.

Reconstruction of the Paper:

In sections 3 and 4, Ekstrom proposes and defends her event-causal indeterminism account of free will (New View).

Ekstrom’s Response to the Regress Problem:1

Ekstrom addresses the Regress Problem in the context of free will. If every free action relies on a prior free action for justification, we run into an infinite chain of actions needing justification. Similarly, we have the same issue if preference formation relies on a prior free preference formation for justification. In Ekstrom’s old view, a decision to act freely is self-determined by the agent’s reasons. The formation of this preference can be causally determined by prior deliberation and reasons; it is still seen as up to the agent since it was formed non-deterministically and without coercive influence. However, the problem arises when considering the formation of the first preference. If this formation is to be free, it cannot be preceded by another preference formation, or else we fall into the infinite regress.

To accommodate the challenge, she proposes a new account that widens the scope of agential cause. In her new account, directly free actions are caused non-deviantly and non-deterministically by agent-involved events (like preferences, desires, and beliefs). This kind of causal history is the basis for free self-determination. Moreover, she claims that there are alternative possibilities for freedom because when the agent performed the act, the agent could have done otherwise since she could have formed a different preference. Therefore, she concludes that her new account is that an action can be self-determined and meet the alternate possibility condition without falling into the regress problem by widening the scope of agential cause. This includes a variety of agential reasons that precede preference formation.

Ekstrom’s Response to the “Disappearing Agent” criticism:2

The author defends against the “Disappearing agent” criticism. This criticism suggests that the agent, on the event-causal indeterminism view, is missing or absent when deciding. Pereboom contends that an agent subscribes to event-causal indeterminism and lacks control over the final decision that emerges from her contemplation. If one finds themselves with reasons supporting divergent choices, the result is not within their control but instead probabilistic. Consequently, chance is responsible for the outcome; in this aspect, her presence or agency is absent.

The author counters this by providing an example of a writer deciding how to spend her afternoon. The writer faces a conflict between professional obligations and personal commitments. If she chooses to work on her paper, it’s not a random event; it’s a decision influenced by reasons such as meeting deadlines and maintaining her professional identity. Reasons are intentional and purposeful, flowing from her freely developed preferences with a causal history. Therefore, actions are not random, nor is it an act appearing from nowhere. This argument is intended to counter the “disappearing agent” because the reasons for actions are part of the agent’s causal contribution to the decision. The author aims to show that the agents remain central to the decision-making process, with their reasons and intentions playing an active and causal role, even if the process is indeterministic.

A Critical Note:

I have a couple of remarks, given that my interpretation of the author’s points is correct. The premise that the preference formation is indeterministic needs to be sufficiently justified. However, the paper is addressed to other libertarian thinkers, so I will assume the truth of the premise here. Given the disclosures, I will provide a critical note.

While it is unclear when children develop control over their preferences and values, there must be a tipping point. Before the tipping point, social influences often have molded a child’s preference; thus, there is a standing preference at the tipping point. How does this initial preference impact a child’s continual preference? If there is no control prior to the tipping point, preference formation seems to be determined up to a point and influences the courses of subsequent preference. Ekstrom may reply that in the length of one’s life, preference will be under the agent’s control.

However, suppose that is the case, but if the probability-raising feature is at play, does preference formation not fail sometimes? In other words, is there an agent disappearing problem at the preference-forming level? Probability-raising in preference formation inherently implies the possibility of failure. If forming a preference is influenced by probabilities, then there is a chance that the preference might not form as intended or might be formed under influences that are not entirely within the agent’s control. While Ekstrom has widened the causal history by including agential reasons encompassing preference, agential reasons face the same indeterminacy. Ekstrom states, “When made for and caused by reasons of the agent’s in deciding, intentions are not passive states but are actively formed. And intention formations that were indeterministically caused by the agent’s formation of certain preferences would not be accidental or lucky (2000, p. 105)” (Ekstrom 2019, 136). I doubt that her point has been sufficiently defended.

An element of luck is involved in preference formation that is characteristically different from action failure. I include values, character, and volitions into preference. The preceding is important in how one responds to reason. Both actions, in a physical sense, and preference construction are guided by the determination of the will. Suppose a person freely acts, and the act is unfruitful; it seems she will be free regardless of the consequences that follow her action. However, when one determines to steer one’s preference or character in a direction but repeatedly fails to do so, or the consequences affect one differently, something randomly affects her preference, character, or value. In other words, the success or failure of preference implementation and formation seems to matter because while the initial determination of the will may be free, the consequences that follow, subsequently affecting one’s will, are not in the agent’s control.

In order to overcome the objection, one must be psychologically resilient to external factors and consequences that follow from one’s action or preference construction. The success or failure of preference formation is consistent with indeterminism. The problem is that part of the indeterminism is random, outside of the agent’s control. Therefore, the conjunction of initial preference staging, for instance, by parents, and preference formation failures should contribute to preference formation that is random, at least partially, to the agent. It is dubious that preference formation is sufficiently under agent control.

I must provide an example of preference formation or expression failure:

Maya was raised in a family with strong ethical values emphasizing honesty. Her parents reinforced the importance of truthfulness and ethical behavior. Growing up, Maya internalized these values, viewing them as central to her identity. As Maya enters the professional world, she lands a job in a corporate environment where the prevailing culture subtly encourages cutting corners and occasionally being less than honest to achieve business goals. Maya initially tries to uphold her values of honesty and integrity in her career; she avoids engaging in dishonest practices and speaks out against unethical actions. However, she realizes that her stance is frowned upon and negatively impacts her career progression; her colleagues with low ethical standards advance more quickly. For prudential reasons, Maya starts questioning the practicality of her values in her current environment. Despite her initial resistance, she gradually conforms to the company's culture, justifying dishonest acts and eroding her values. She starts rationalizing actions that she would have considered unethical before, indicating a failure to maintain her original value system in the face of external pressures. Still, in an alternate scenario, Maya is invited to apply for another corporation with a different culture and higher ethical standards. She applies, transitions, and solves her problems without compromising her values.

In theory, Maya had an alternate possibility; she could have quit her job. Additionally, one may argue that Maya’s true preference was career success in the first place. This raises an epistemic problem where we may only know our preference by hindsight. However, the problem of interest here is about the role of luck in preference formation and maintenance. As seen in the alternate scenario, the unexpected job offers at an ethical company prevented the erosion of Maya’s certain values. In other words, it made maintaining her values relatively easy, provided honesty is her preferred value due to luck. Suppose Maya chooses to transfer companies. Can she do it with self-direction, non-deviant causation, absence of compulsion, and alternative possibilities? I doubt that she can be free from non-deviant causation.

Ekstrom may contend that freedom consists in responding to external factors. However, maintaining one’s preference encompassing character and values must be important. The external factors penetrate one’s psychological state, introducing reasons out of nowhere and shaping character and preference. Agential control is how an agent responds to external factors and how one actively preserves or modifies one’s internal states considering external influences. As shown in the example, certain, external factors influence one without agential permission, randomly. Even if indeterminism is true regarding preference formation in the like, some are within one’s control, but others are random. The desideratum for libertarians is to argue that the amount of control one has is sufficient for agent control.

I believe compatibilism is the only viable option; free will merely amounts to choosing the best option one thinks, within constraints, among alternate options. While there are alternate options, the thinker is determined to choose what she thinks best, based on information (including mental state) available to them. Similarly, she determinately chooses a preference and values certain reasons, ends, and means based on available (and accessible) information. Of course, this preliminary view must be further developed. My view implies that a great amount of moral responsibility lies in the society that provides the foundation for developing moral agents. Effective moral agents require sufficient resources and well-being. Contra libertarianism, it demands social cooperation, not individualism.

An act is free even if one fails to act according to their actively formed preference with a causal history as long as the agent possesses control over developing that preference on Ekstrom’s account (Ekstrom 2019, 136). I am unpersuaded that an agent can construct preference with sufficient control under indeterminism. It is better to embrace determinism in the preference formation stage and argue for a compatibilist approach to free will or sufficiently explain how one can be in control under indeterminism.


  1. Clarke holds that the active formation of preference cannot be a free act on pain of regress (Ekstrom 2019, 135). ↩︎
  2. Pareboom (2014), for instance, is interested in the disappearing agent problem
    concerning actions; he argues for agent-causal libertarianism. ↩︎


  • Ekstrom, 2019. “Toward a Plausible Event-Causal Indeterminist Account of Free Will,” Synthese, 196: 127–144.
  • Mele, Alfred R., 1995. Autonomous Agents: From Self-Control to Autonomy, New York: Oxford University Press.
  • –––, 1996. “Soft Libertarianism and Frankfurt-Style Scenarios,” Philosophical Topics, 24(2): 123–41.
  • –––, 1999b. “Ultimate Responsibility and Dumb Luck,” Social Philosophy & Policy, 16: 274–93.
  • –––, 2006. Free Will and Luck, New York: Oxford University Press.
  • Pereboom, Derk. “The Disappearing Agent Objection to Event-Causal Libertarianism.” Philosophical Studies 169, no. 1 (2014): 59–69.

*Disclaimer: This article is a creative interpretation and synthesis work, drawing inspiration from multiple sources. While efforts have been made to ensure the accuracy of the interpretation or information presented, readers are encouraged to consult the source material for a comprehensive understanding.

**Please note that I often write the Critical Note (constructive criticism) section after posting the article. If the section is left blank, it is intentional.

I use the term “Critical” in a broad academic sense, encompassing thorough examination, balanced evaluation, and thoughtful interpretation. It is not limited to pointing out flaws or offering negative criticism.

**I reserve the right to edit this page at my convenience.