As an amateur poker player (very amateur), I enjoy watching Youtube videos showing top poker professionals practice their craft. Watching the pros play cards, and the reading the comments left by other watchers, can be very instructive. Different viewers have radically different reactions to the play of various poker hands (Was it brilliant? Or was it insane?), and those differences reflect the different levels of poker savvy those viewers have. A lot of what happens between top poker players will simply fly over the head of a complete newcomer to the game. Unless you have a certain level of skill, you won’t understand what is taking place, and why one play of a hand counts as brilliant while another qualifies as a disastrous miscalculation.

Reading Rational and Social Agency: The Philosophy of Michael Bratman reminded me a lot of those poker videos. The book collects a number of papers reflecting on the work of Michael Bratman, one of the leading philosophers of action in the world today. The papers engage with Bratman’s ideas at a very high level. But as with a table full of poker pros, the level of play on display is very high. And so if you don’t know the game very well, you won’t get much out of it. But if you know enough about the game to recognize high-level play when you see it–even if you cannot play at that level yourself–you will find the whole experience quite rewarding.

The book’s introduction, by editors Gideon Yaffe and Manual Vargas (chapter 1), provides a very readable, but very sophisticated, introduction to Bratman’s major contribution to the philosophy of action, the planning theory of intention. This introduction nicely associates the planning theory with Herbert Simon’s theory of bounded rationality, and contrasts both to the traditional rational choice model of decision-making. According to this traditional model, agents have preferences and beliefs, which together uniquely dictate the (utility-maximizing) course of action to the agent. Moreover, agents update their beliefs as they receive new information (like good Bayesians), and then change how they act accordingly. Bratman, like Simon, argues that the traditional rational choice model simply does not accurately describe human behavior. It also does not work as a prescriptive model; human beings deciding how to act simply shouldn’t try to follow the dictates of the rational choice model, as its ideals are simply unrealistic to agents like us (p. 3). Instead, Bratman argues that human beings rely, and should rely, on plans. These plans coordinate their behavior over time and with other human beings. These plans do not uniquely dictate courses of action, but they restrain decision-making in ways that make it manageable to mere mortals (not the effortlessly logical agents that populate rational choice models). Moreover, these plans are not immediately revised as circumstances change or new information obtained; there is a threshold that must be crossed before it becomes worthwhile for people to rethink the plans they have made.

There are many ways that one could respond to Bratman’s planning theory. Rational and Social Agency collects ten such responses, plus the introduction and a lengthy concluding essay by Bratman himself. Richard Holton’s “Intention as Model for Belief” (chapter 2) asks whether beliefs function for human beings in a manner similar to Bratman’s plans. Just as one does not immediately revise one’s plans just because one has received some small amount of information, so one does not immediately revise one’s beliefs. The Bayesian model of belief formation is thus unrealistic for human agents. Alfred Mele’s “The Single Phenomenon View and Experimental Philosophy” (chapter 3), concerns itself with a problem that may strike some readers as obscure–can an agent intentionally do A without intending to A? These two papers are relatively standalone; they do not seem closely connected either with each other or with the other papers in the volume. With the other papers, however, it is easier to identify various types of connection.

For example, both Kieran Setiya’s “Intention, Plans, and Ethical Rationalism” (chapter 4) and J. David Velleman’s “What Good Is a Will?” (chapter 5) use G.E.M. Anscombe’s philosophy to propose radical modifications to traditional understandings of human action. Setiya suggests that we should be prepared to abandon the idea that human beings can intend to A but fail to A. Velleman would reject the idea that intentions serve practical purposes, like coordinating our actions over time and with other people; instead, he suggests that they simply help human beings to understand what they themselves are doing. Both R. Jay Wallace’s “Reason, Policies, and the Real Self: Bratman on Identification” (chapter 6) and “Desires…and Beliefs…of One’s Own,” by Geoffrey Sayre-McCord and Michael Smith (chapter 7), consider the problem of identifying which of my beliefs, desires, and other cognitive states count as mine, as opposed to something that just happens to me. Wallace does not believe that Bratman can make this distinction without normative judgment; we do not count beliefs, desires, etc. as our own unless we have also judged them right, true, good, or otherwise desirable. Sayre-McCord and Smith, by contrast, would readily attach to us desires and beliefs that we do not like; the alcoholic might wish to be rid of her constant craving for liquor, but the craving should nonetheless count as part of her identity.

Wallace and Sayre-McCord and Smith thus challenge Bratman from opposite directions. Similar opposing challenges can be found in the last four critical papers in the volume. Elijah Millgram’s “Segmented Agency” (chapter 8) suggests that planning à la Bratman may be rudely interrupted by dramatic changes of circumstance. The resulting change of plans may be so extreme as to call into question the idea that a single agent is responsible for it all. Christine Korsgaard, in “The Normative Constitution of Agency” (chapter 9), very much wishes for agency to be unified, and believes that Bratman’s account cannot do this in the right way. Korsgaard believes that agency essentially involves an active project of self-construction on the part of the agent. Chapters 10 and 11 both deal with the problem of shared agency, again critiquing Bratman from radically different directions. Margaret Gilbert holds in “The Nature of Agreements: A Solution to Some Puzzles about Claim-Rights and Joint Intention” that shared agency requires something very much like agreement, with an associated set of rights and obligations. This is much stronger than what Bratman would require for shared agency. Scott Shapiro would require something much weaker. In “Massively Shared Agency,” he argues that the world contains shared agents comprised of enormous numbers of people, many of whom do not even know each other, and none of whom need care whether or not their coordination with others succeed. (“I just work here.”) All that these shared agents really need is commitment to a shared plan (p. 280).

Some of these challenges to Bratman strike me as easy to meet. Sayre-McCord and Smith, for example, are no doubt correct that there is some sense in which the desires that an agent rejects count as her own. (Who else could claim them?) But this sense strikes me as rather trivial, and any interesting sense of agency is going to require something more robust, whether it be Bratman’s account or some other. Other challenges strike me as posing interesting challenges meriting further investigation. I was particularly intrigued by Holton’s challenge to Bratman on the problem of belief. Holton holds that the states of the world I regard as possible vary with the problems I confront. This means that my beliefs change with my practical needs. Bratman, however, does not want beliefs to be determined by anything other than theoretical considerations (good evidence, etc.), and so he draws a distinction between believing something to be true and taking something for granted in a particular decision-making situation (pp. 309-310). I am not sure who has the right approach here, but if Bratman’s challenge to the traditional rational choice model is justified, then working out all the necessary changes to our understanding of belief, desire, and so forth would be a critically important next step.

It is very difficult to do justice to the conversation taking place within Rational and Social Agency between Bratman and his critics. It is no small commitment to undertake reading this book. (One might even say it requires committing to a plan.) The reader with a certain level of knowledge about the philosophy of action will learn a lot about a compelling theory of human agency. This reader will also be moved to think more generally about what purposes a theory of agency is supposed to serve, what questions it must answer, and what challenges it must overcome. The experienced reader will thus gain a lot from this book. But I don’t recommend a complete newcomer start with Rational and Social Agency. Playing heads-up no-limit Texas Holdem against Phil Ivey might prove easier.

© 2015 Peter Stone

Dr. Peter Stone, Ussher Assistant Professor of Political Science, Trinity College, Dublin, Ireland