Act Individuation: An Experimental Approach
- First Online:
- Cite this article as:
- Ulatowski, J. Rev.Phil.Psych. (2012) 3: 249. doi:10.1007/s13164-012-0096-1
- 103 Views
Accounts of act individuation have attempted to capture peoples’ pre-theoretic intuitions. Donald Davidson has argued that a multitude of action descriptions designate only one act, while Alvin Goldman has averred that each action description refers to a distinct act. Following on recent empirical studies, I subject these accounts of act individuation to experimentation. The data indicate that people distinguish between actions differently depending upon the moral valence of the outcomes. Thus, the assumption that a single account of act individuation applies invariantly seems mistaken.
Suppose that Tiffany moves her arm, depresses the lever, lifts the weight, operates the Nautilus machine, and scares the man on the rowing machine all at the same time. Do the descriptions “Tiffany’s moving her arm” and “Tiffany’s depressing the lever” refer to distinct acts? Or do they refer to the same act? Our intuitions tell us that actions have boundaries, though the boundaries between them may not be clear. Determining what feature distinguishes one act from others has been called the problem of act individuation. The problem of act individuation has been largely concerned with descriptive features of action, what causes x to occur and what x causes to occur. So, given that the problem is descriptive, it seems correct to believe that our distinguishing between acts are empirically discoverable. My paper’s goal is to redirect (as well as to reinvigorate) this debate through the employment of empirical methods.
1 The Problem of Act Individuation
At least two accounts have seemed intuitively plausible responses to the problem of act individuation. The first is the minimizing view, and the second is the maximizing view. On the minimizing view endorsed by Donald Davidson (1963), Tiffany performs one act which may be described in various ways. According to the maximizing account of Alvin Goldman (1970, 1971), for each and every description of Tiffany’s act there is a corresponding act with which it is associated.1 Both the minimizing and maximizing views have aimed at devising an account that is consistent with the ordinary person’s pre-theoretic conception of action. Proponents of each view have said that whenever a person is presented with a case, like the Tiffany case I present above, the person will distinguish acts according to their own account. I call this the “assumption of invariant individuation.” The problem is that neither view has considered whether the moral valence of the consequences of an action might infect the ordinary person’s view of individuating action. If it can be shown that moral appraisals play a role in how ordinary people distinguish between acts, then the assumption of invariant individuation the minimizing and maximizing view uphold might be mistaken. In later sections of this paper, I will reveal some empirical data that call into question the assumption of invariant individuation. In this section, I will review the maximizing and minimizing views of act individuation and show how Joshua Knobe’s pioneering work motivated my undertaking an experimental study of act individuation.
In spelling out his version of the minimizing view, Davidson provides a criterion for the identity of events: “events are identical if and only if they have exactly the same causes and the same effects” (Davidson 1969, 177ff). Since an action bottoms out in bodily movement (or what Davidson calls “primitive action”), we “never do more than move our bodies: the rest is up to nature” (Davidson 1971, 59). A bodily movement gives rise to a chain of events. Each event allows for a different description of action. Thus, for each bodily movement, we have a number of descriptions that satisfactorily describe that single act.
I flip the switch, turn on the light, and illuminate the room. Unbeknownst to me I also alert a prowler to the fact that I am home. Here I need not have done four things, but only one, of which four descriptions have been given. (Davidson 1963, p. 4)
We might not call my unintentional alerting of the prowler an action, but it should not be inferred from this that alerting the prowler is therefore something different from flipping the switch, say just its consequence. Actions, performances, and events not involving intention are alike in that they are often referred to or defined partly in terms of some terminal stage, outcome, or consequence. (Ibid.)
Applying Davidson’s criteria to the “the Tiffany case” yields that moving her arm, depressing the lever, lifting the weight, operating the Nautilus machine, and scaring the man on the rowing machine all refer to a single act of Tiffany’s. Tiffany’s moving her body in just the way described has all of the following effects: moving her arms, depressing the lever, and scaring the man on the rowing machine. Tiffany needs to do nothing more than move her body for all of these effects to occur.
An act-type, as I construe it, is simply an act property, something that an agent exemplifies. When we say, “John weighed 170 lb” or “John was bald,” we ascribe to John the property of weighing 170 lb or the property of being bald. Similarly, I suggest that when we say “John signaled for a turn” or “John killed George,” we ascribe act properties or act types to John: the property of signaling for a turn or the property of killing George. (Goldman 1971, 769)
To perform an act, then is to exemplify a property. To perform the act of giving a lecture is to exemplify the property of giving a lecture. A particular act, then consists in the exemplifying of an act-property by an agent at a particular time. I shall call such particular acts: “act-tokens.” An act-token is not itself a property. It is the exemplifying of a property by an agent. (Goldman 1970, 10)
Since an act-token is the exemplifying of a property by an agent at a time, it is natural so to individuate act-tokens that two act-tokens are identical if and only if they involve the same agent, the same property, and the same time. (Ibid.)
Suppose there is an action x that has the property P at time t1. On Goldman’s maximizing account, if there is some action y that has property Q at time t*, then x and y are the same if and only if x=y, P=Q, and t1=t*. Otherwise, x and y are distinct act-tokens that exemplify distinct act-types.
If we apply Goldman’s criteria to the “the Tiffany case,” moving her arm, depressing the lever, lifting the weight, operating the Nautilus machine, and scaring the man on the rowing machine are distinct because each act-token exemplifies a different property. Moreover, adding an adverbial modifier to a description, such as “Tiffany’s lifting the weight slowly,” according to Goldman, will render it distinct from a description that does not include an adverbial modifier, such as “Tiffany’s lifting the weight.” Although the acts are distinct on Goldman’s account, he contends there is a certain relationship between them. The relationship between them is not one of identity.3 If it were identity, Goldman’s view of act individuation would collapse into the minimizing account.
One explicit aim of Goldman’s project is “to explicate certain aspects of our common sense conceptual scheme… [and it] is intended to capture, as closely as possible, our pre-theoretic conception” of action (Goldman 1970, p. vi). The goal of Goldman’s project is to articulate the folk’s ordinary conception of action, but his account seems merely to be a reflection of what he, a specialist, thinks the folk’s intuitions are. I say this because Goldman cites no evidence and also offers no empirical evidence of his own supporting the view that his understanding of the people’s pre-theoretic conception corresponds with his own view.
[E]mpirical studies are profoundly important in helping us (as theorists) understand the ordinary grasp of moral matters, which is a part of the task of doing metaphysics in the ethical realm … [T]he ways we conceptualize many types of entities of metaphysical interest are not available to introspection. … We need the help of cognitive science to illuminate these conceptualizations. (Goldman 2007, 465)
According to Goldman, our attention to cognitive science gives rise to an ordinary folk pretheoretic view of individuating actions and events. Goldman uses the currently existing developmental psychology literature on individuating objects to show that the mind might employ two different systems of representing events. The spatio-temporal system represents actions in terms of bodily movements, and so seems to be consistent with Davidson’s unifier view. The second “kinds” system assigns a primary role to the property or kind a token action instantiates, and so is consistent with Goldman’s maximizing account. Instead of positing categories of events based upon the flourishing developmental psychology literature, I have taken Goldman’s recommendation seriously and performed a series of experimental studies on individuating action. I believe what I have found partially supports a two systems view, though in this paper I will leave aside whether what I have found does support the view.
Like Goldman, Davidson’s view is meant to capture people’s pre-theoretic conception of act individuation. Davidson sets out “to defend the ancient—and common sense—position” (Davidson 1963, 3). We may think that the purpose of Davidson’s account of act individuation is to offer a causal explanation of technical concepts, such as human agency and action. But, in the Introduction of Essays on Action and Events, Davidson tells us “the ordinary notion of cause which enters into scientific or common sense accounts of non-psychological affairs is essential also to the understanding of” human action (Davidson 2001, xv).
For both Goldman and Davidson, a theory of act individuation gives fixed and specific criteria for how to distinguish between acts. These criteria should reflect what the ordinary or common sense view is. Neither Goldman nor Davidson ask people what their views are. Unbeknownst to them, the criteria of act individuation may be affected by some other important factors, such as the moral consequences of some action. Their assumption that the criteria of individuation do not vary may be called into question if it is discovered that the common sense view does not coincide with their own.
Elsewhere in action theory, Joshua Knobe (2003) has been exploring the ordinary conception of intentional action. His results have been surprising! Knobe discovered that a majority of subjects judged that an agent intentionally brought about a side effect when the side effect was bad. What makes his findings surprising is that subjects did not believe a side effect was brought about intentionally when that side effect was good. Not everyone agrees with Knobe’s explanation that the moral goodness or badness of the side effect drives people’s intentionality judgments.4 These experimental studies have shared a single purpose in showing that the ordinary conception is very different from what philosophers have taken it to be. Knobe’s methodology gives us a way of exploring people’s conceptions of act individuation to show whether Davidson and Goldman’s assumption about people is correct. Knobe’s findings suggest a hypothesis: act individuation will be sensitive to the valence of the consequences of the action (just as judgments of “intentionally” are).
The vice-president of a company went to the chairman of the board and said, ‘We are thinking of starting a new program. It will help us increase profits, but it help the environment.’/help the environment.’
The chairman of the board answered, ‘I don’t care at all about harming/helping the environment. I just want to make as much profit as I can. Let’s start the new program.’
They started the new program. Sure enough, the environment was harmed/helped.
According to Knobe’s findings, when the chairman’s action brings about harm to the environment, people judge that the chairman does so intentionally. Conversely, people judge that the chairman’s helping the environment is not intentional. The asymmetric result is unusual, and numerous explanations of the effect have been given. We must remember that the chairman’s harming or helping the environment is merely a side effect of the chairman’s starting a new program. To my mind, too much emphasis has been placed on the intentionality of the chairman’s acts. Perhaps it is the case that people’s intentionality judgments slip between the two acts because they may not be distinguishing between the chairman’s “starting the new program” and “harming the environment.” People may not believe that two acts are distinct because of the consequence’s moral valence. Knobe’s experiments seem to suggest there is something going on further upstream from the studies of intentional action concerning how people individuate actions.
Using the Knobe experiment as a prototype, I designed a series of experiments that would yield results showing us how the ordinary person distinguishes between acts. The first experiment included two cases. The cases asked respondents to indicate how many acts the agent appearing the vignette performed. My hypothesis for this experiment was that participants would judge that the agent performed fewer acts based upon the moral valence of the consequences of the action. Because enumerating actions is not equivalent to identifying actions, in the second experiment, I decided not to ask participants to count the number of actions but whether they distinguished between supererogatory actions and ordinary actions. The trouble with asking people to distinguish between supererogatory and an ordinary property of action is that it is difficult to know that people are tracking adverbial modification in the way we would expect them to when asking about individuating action. So this worry gave rise to the third and fourth experiment. To my mind, the results of these two experiments get us closest to the ordinary conception of act individuation. For the third and fourth experiment, I expected them to generate an asymmetry similar to what Knobe found in his experiment. My hypothesis is that subjects’ views of the valence of the consequences of an agent’s actions will influence whether they believe two or more descriptions designate the same or distinct acts in the manner we are considering. Consequently, we may want to say that the search for an account of act individuation that applies invariantly like those proposed by Goldman and Davidson may well fail.
2 The First Experiment5
The first experiment asks respondents to count the number of acts an agent performed. My prediction is that subjects’ moral appraisals will affect their judgments about individuating action. Subjects will say that an agent performs fewer acts based upon the moral valence of an act’s consequences.
To test my hypothesis, I conducted an experiment. First, I designed three vignettes that were like Alvin Goldman’s examples, which he used to undermine the minimizing account of act individuation (Goldman 1970, 2ff). Subjects were 26 undergraduate students in an introductory philosophy class at the University of Utah. A study proctor distributed the surveys to subjects, and subjects were recruited from introductory philosophy classes not taught by the author. Each subject received all three of the following cases.6
Case 1: Suppose that John is playing the piano, and that his playing the piano causes Smith to fall asleep while also causing Brown, who was already asleep, to wake up.
Case 2: Suppose that Smith moves his arm, operates the pump, and replenishes the house’s water-supply. Unbeknownst to Smith, the water is poisoned. So, Smith poisons the inhabitants of the house.
For Case 1, subjects were asked whether John performed three distinct acts or one act, and, similarly, for Case 2, subjects were asked whether Smith performed four distinct acts or one act.
The prediction was successful. In Case 1, a majority of subjects (89 %) responded that John performed one act, and, in Case 2, a majority of subjects (85 %) said that Smith performed one act. For Case 1, the difference was statistically significant χ2 (1, N = 26) 15.385, p < .0001, and, for Case 2, the difference was statistically significant χ2 (1, N = 26) 3.846, p < .05.
There are at least two problems with cases 1 and 2.7 First, one could say of case 1 that there is no negative consequence, so I should have predicted that participants’ moral appraisals would not affect their responses.8 Although I agree that the case is ambiguous and could lead someone to make such a claim, I believe it is possible that people’s moral judgments do play a role in enumerating the actions undertaken by John. People may focus on John’s waking up Brown, and they may consider that a bad thing. Anyone who has been awoken unexpectedly by a piano may believe it very bad to have been stirred from sleep that way. Knowing this about themselves people would think because John’s waking up Brown refers to the same act as John’s playing the piano, I also should not distinguish John’s piano playing from John’s putting Smith to sleep. Participants believe John’s waking up Brown is bad, and this leads people to say that John performed just one act.
Second, a critic may contend that cases 1 and 2 might fail to provide support for my hypothesis because both conditions are counting cases. The cases ask respondents to identify how many acts the agent performed, whereas questions about the individuation of action should ask respondents whether some act is identical to another act. Some kinds of things can be individuated but not enumerated (Wiggins 2001, 60f). Enumerating actions is distinct from the identity of actions (Jones 1964, 201ff). So, questions about the number of acts and questions about the identities of acts are distinct. Because of these two problems with cases 1 and 2, a second experiment was designed to ask participants whether they distinguished between ordinary acts and supererogatory ones.
3 The Second Experiment
Those participants who received Cases 1 and 2 were also given the following case. Cases 1 and 2 asked respondents to count the number of actions. Counting the number of actions an agent performs seems to fail in distinguishing between actions. So, in the second experiment, I asked respondents whether they distinguish between supererogatory and ordinary properties. The act in the vignette either was an ordinary act or a supererogatory act, one that agent could do but does not have an obligation to do. Asking respondents whether they thought it was “especially nice” of the agent to perform a supererogatory act would permit me to say that a supererogatory act and an ordinary act are distinct. If participants respond negatively, that it was not “especially nice” of the agent to perform a supererogatory act, then it is possible that people fail to distinguish between supererogatory acts and ordinary acts. If participants respond positively, that it was “especially nice” of the agent to perform a supererogatory act, then it is possible that people distinguish between supererogatory acts and ordinary acts.
Goldman has argued that two putatively identical acts fail to coexemplify a given property in supererogatory cases. Just as for Cases 1 and 2, my prediction is that people’s moral judgments play a role in how they individuate acts. For supererogatory cases, people will distinguish between an act being especially nice and an ordinary act, i.e. repayment of a debt, and will fail to distinguish between an act being especially nice and a supererogatory act, i.e., repayment of a debt using a unique form of currency. People’s individuating action corresponds largely with Goldman’s recent view that kinds play a critical role in the ordinary conceptualization of action. He writes, “According to this approach, if action a is a token of action-type A and action b is a token of action-type B, then a is not identical to b, even if a and b are performed by the same agent at the same time” (Goldman 2007, 472). People’s conceptions will correspond, perhaps roughly, with the property-exemplification view Goldman discussed in his early work.
Suppose that Donna owes Stacey two dollars. Seeing Stacey on the street, Donna reaches into her pocket for some cash. She discovers two single dollar bills and one two-dollar bill. Donna likes to collect two-dollar bills herself, but she recalls that Stacey simply goes wild over them.
Bearing this in mind, Donna pays Stacey the money with the two-dollar bill.
After reading this vignette, subjects were asked two questions, which were counterbalanced for order effects: (a) “Was it especially nice of Donna to payback Stacey with two dollars?” and (b) “Was it especially nice of Donna to payback Stacey the two dollars with the two-dollar bill?” Subjects could respond by circling either “YES” or “NO.” As predicted, subjects gave asymmetric responses. They distinguish between properties that are supererogatory (namely, Donna’s giving Stacey the two-dollar bill) and those properties that are not supererogatory (namely, Donna’s paying back Stacey two dollars). A majority of subjects (77 %) responded that it was not especially nice of Donna to payback Stacey the two dollars, and a majority of subjects (96 %) said that it was especially nice of Donna to payback Stacey with the two-dollar bill. The difference was statistically significant χ2 (1, N = 26) 28.836, p < .0001.
Whereas people believed it was not especially nice of Donna to pay back Stacey with two dollars, they judged it was especially nice of Donna to pay back Stacey the two dollars with the two-dollar bill. So, participants seem not to distinguish between Donna’s paying back Stacey the two dollars and Donna’s paying back Stacey with two dollars, and they do seem to distinguish between Donna’s paying back Stacey the two dollars and Donna’s paying back Stacey with a two-dollar bill. The questions that followed the vignette may not give us what the ordinary conception of act individuation is because the questions did not ask participants whether they believed the two action descriptions refer to the same act. The questions concerned whether it was especially nice of Donna to pay back Stacey one way or another and not individuating action. Given that the second experiment may not have uncovered something about people’s views of act individuation, I devised a third experiment that tested whether people believed the two descriptions refer to the same act or not.
4 The Third Experiment
Smith’s job is to pump water into the cistern which supplies the water of a house.
One day Smith operates the pump and replenishes the house’s water supply. The occupants of the house are healthy and have no health problems. Jones tells Smith that someone has found a way of systematically contaminating the water’s source with a deadly cumulative poison whose effects are unnoticeable until they can no longer be cured.
Smith says, “I don’t care about contaminating the water’s source; I just want to earn my pay.”
The occupants of the house drink the water. Sure enough, they are poisoned and die.
Subjects receiving the poison condition were then asked: “Was Smith’s operating the pump the same thing as his poisoning the house’s inhabitants or were they distinct?”9 Subjects were able to choose either “YES” or “NO.”
Smith’s job is to pump water into the cistern which supplies the water of a house.
One day Smith operates the pump and replenishes the house’s water supply. The occupants of the house are sick and have severe infections. Jones tells Smith that someone has found a way of systematically purifying the water’s source with a cumulative antibiotic whose effects are unnoticeable until they cure someone who has a severe infection.
Smith says, “I don’t care about purifying the water’s source; I just want to earn my pay.”
The occupants of the house drink the water. Sure enough, they are saved and live.
Subjects receiving the savior condition were then asked: “Was Smith’s operating the pump the same thing as his saving the house’s inhabitants or were they distinct?”
Each condition elicited different patterns of responses. In the poison condition, most subjects (61 %) said that Smith’s operating the pump and his poisoning the occupants were the same thing, whereas in the savior condition, most subjects (87 %) said that Smith’s operating the pump was distinct from his saving the inhabitants. The difference was statistically significant, χ2 (1, N = 75) = 17.613, p < .001.
An asymmetric pattern of responses occurs in the results of the two cases, even though very little distinguishes them. It seems that in each case some feature of Smith’s action affects people’s intuitions. The question is: why would people respond to the two vignettes about action individuation differently?
In the poison condition, people are more prone to say that each description designates the same act. People seem to believe that Smith’s operating the pump is the same thing as his poisoning the house’s inhabitants. It appears that people believe the two descriptions designate the same act when a negative consequence is brought about by Smith’s action. Although Smith is not operating the pump in order to harm the house’s inhabitants, people’s intuitions support the view that the two descriptions of Smith’s activities designate the same action.
In the savior condition, people are far less inclined to say that each description designates the same act. When Smith’s operating the pump brings about a positive consequence, people are reluctant to say that the two descriptions designate the same act. Smith is not operating the pump in order to save the house’s inhabitants. The results are asymmetric because, in the savior condition, a majority of people believe each description designates distinct acts, while, in the poison condition, a majority of people believe that both descriptions refer to the same act.
5 The Fourth Experiment
Someone may complain that subtle differences between the poison and savior conditions in the second experiment have an overwhelming influence of eliciting asymmetric responses. For example, in the poison case, the “effects are unnoticeable until they can no longer be cured,” but in the savior case, the “effects are unnoticeable until they cure someone who has a severe infection.” The poison causes the illness, but for the savior case the antibiotic is curing something totally distinct from it, namely, an infection. The two conditions are not equivalent because pumping a poison into a house and it poisoning the people is closer than pumping an antibiotic into a house and that curing someone. Given this possible complaint, I devised a fourth experiment.
The vice-president of a company went to the chairman of the board and said, “We are thinking of starting a new program. It will help us increase profits, and it will also harm the environment.
The chairman of the board answered, “I don’t care at all about harming the environment. I just want to make as much profit as I can. Let’s start the new program.”
They started the new program. Sure enough, the environment was harmed.
Subjects receiving the harm condition were asked: “Is the chairman’s starting the new program the same thing as the chairman harming the environment?” Subjects were able to choose “YES” or “NO.”
The vice-president of a company went to the chairman of the board and said, “We are thinking of starting a new program. It will help us increase profits, and it will also help the environment.
The chairman of the board answered, “I don’t care at all about helping the environment. I just want to make as much profit as I can. Let’s start the new program.”
They started the new program. Sure enough, the environment was helped.
Subjects receiving the help condition were asked: “Is the chairman’s starting the new program the same thing as the chairman helping the environment?” Subjects were able to choose “YES” or “NO.” As predicted, a majority of respondents (84 %) said that the chairman’s starting the new program was the same thing as harming the environment, and, in the help case, a majority of respondents (60 %) said that the chairman’s starting the new program was not the same thing as helping the environment. The difference was statistically significant, χ2 (1, N = 78) = 16.028, p < .001.
What these results suggest is that the valence of the consequences of an action plays a role in how we distinguish between actions. The assumption of invariant individuation among action theorists is that a unified account of action individuation will apply to different sorts of cases. For example, Goldman has shown how his simple, unified view of action individuation applies to many different cases (Goldman 1970, pp. 5–10; Goldman 1971, pp. 487–489). But, as the data seem to suggest, it might be that people individuate actions in different ways in different cases. People’s responses might cluster around a central core or theory, but—then again—they might not. The experimental results seemingly overturn the assumption of invariant individuation.
There have been a variety of attempts to explain the underlying cause of Knobe’s now well-known side effect effect. One possibility, which is consistent with the evidence, is that it is due to deeper phenomenon involving asymmetry in act individuation. That is, asymmetry in intention ascription is due to asymmetry in act individuation. If you view act A and act B refer to the same act, then obviously if you say someone intended act A, you have to say the person also intended act B. But if you do not think act A and act B are the same act, then the fact someone intended A need not lead you to think they intended B (or even performed B). More specifically, if you are inclined to treat the negatively valenced act (poisoning) of operating the pump as just the same act as operating the pump, then you are going to think an intention to operate the pump is also an intention to poison. But if you are inclined to treat the positive side effect (curing) of operating the pump as different from operating the pump, then you are going to think an intention to operate the pump is not an intention to cure. Further study would be needed to confirm this speculation.
Of course, none of what I have said here is the last word on the folk conception of act individuation. More empirical work should be completed to shed light on it. Other empirical questions may also be important for further inquiry. That there is an asymmetry present in the data goes to show only that how people distinguish between actions might depend on the valence of the consequences of an action.
Besides Goldman, Kim (1969, 1976) has offered a maximizing account of event individuation. If one believes the prevailing assumption that actions are events, then one can apply Kim’s account to actions. For purposes of this paper, my focus will be Goldman’s maximizing account of act individuation. Perhaps some of my future projects will address Kim’s account of event individuation directly.
Goldman (1970, 1971) presented a maximizing account of act individuation. But his recent work suggests that he has revised the maximizing account of act individuation, based upon his research in developmental psychology and cognitive science (Goldman 2007). It is difficult to say whether Goldman has withdrawn support of his original view because (i) his recent article frames the maximizing account very favorably and (ii) the aim of the recent work is to show that developmental psychology and cognitive science inform debates in metaphysics. This paper presents empirical studies that seem to support Goldman’s two-systems view. I discuss below what the two-systems view is.
For purposes of this paper, I will not summarize the various relationships Goldman says may hold between act-types and act-tokens.
See, e.g., Adams and Steadman 2004a, 2004b; Cushman and Mele 2008; Feltz and Cokely 2007; Hindriks 2008; Leslie et al. 2006; Machery 2008; Malle 2001, 2006; Malle and Knobe 1997; Mallon 2008; McCann 2005; Meeks 2004; Nadelhoffer 2004a, b, c; Nado 2008; Nanay 2010; Nichols and Ulatowski 2007; Phelan and Sarkissian 2008, 2009; Sripada 2010; Turner 2004; Wiland 2007; Wright and Bengson 2009; Young et al. 2006.
Experiment 1, 2, and 3 were completed over a 6-year period at three different institutions, University of Utah, Weber State University, and University of Mississippi. I believe that the data represent a cross-section of the college-aged population, though only in the third experiment did I ask for demographic data.
The conditions were counterbalanced to determine whether any order effects were present. None were detected. I will admit that the first study did not include a sufficient number of subjects to make any overwhelming claims about individuation of action, but the results did give me reason to continue testing subjects.
These two criticisms are in addition to the usual criticism that the subject pool was too small to make any substantive points about action individuation.
I thank an anonymous referee for pointing out this potential problem.
Someone might wonder why I used “thing” rather than “act.” Thing is less loaded than act. I say this and defend it in three ways. First, Jennifer Hornsby (1979) has pointed out that “I did the same act as you” asserts an identity between actions (i.e., the minimizing view), while “I did the same thing as you” does not. Since my experiment tries to discover what ordinary talk might reveal of the identity of actions, I did not want my question to overly burden the multiplier or the componential view. One may even say that if I had used “same act(ion),” then my experiment begged the question against the two other views. Second, many action theorists, such as Goldman (1970), have noticed a relevant distinction between the terms “act” and “action.” In fact McCullagh (1976) has argued that the impasse between the different accounts of individuating actions may be settled by more carefully distinguishing between actions and acts. I did not want to discuss the distinction between act and action or to be compelled to discuss the distinction with subjects, so I decided to use thing instead. Finally, the term “thing” does not necessarily stand in for “objects” or “entity” for ordinary folk as it may for philosophers. Philosophers may think that “thing” refers to an nondescript object, but the folk read “is x the same thing as y?” as we philosophers might read, “is x qualitatively the same as y?” This final reason was based on talking with a few non-philosophers about my experiments. When I asked, “is operating the pump the same as poisoning the inhabitants?” Their responses included: “yeah, they’re the same thing” or “naw, they’re not the same thing.”
I would like to thank Eric Amsel, Adam Feltz, Joshua Knobe, Ron Mallon, Shaun Nichols, Sarah Paul, Bill Ramsey, and Walter Sinnott-Armstrong, as well as audience members at the 2008 MidSouth Philosophy Conference, the 2010 Joint Meeting of the North Carolina Philosophical Society and South Carolina Society for Philosophy, and the 2010 Southern Society for Philosophy and Psychology for helpful suggestions and comments on earlier drafts of this paper. I am also grateful for two anonymous referees employed by the Review of Philosophy and Psychology who were instrumental in helping improve the quality of this paper.