Asymmetries in Folk Judgments of Responsibility and Intentional Action

John Bengson Jennifer Wright

University of Texas at Austin University of Wyoming

A straightforward relation

Whether or not a given action is intentional often matters to us when we assess the goodness/badness of the action and the responsibility (e.g., praise/blame) of the actor.

If x acts to bring about an outcome O intentionally, then we may form a different judgment about x or x’s action than if O is simply an accident or the result of (non-willful) ignorance.

Intentionality –influence Evaluative attributions judgments

A surprising relation

Yet, recent work in “experimental philosophy” suggests a bi-directional relation between intentionality attributions and evaluative judgments in the folk.

(Knobe, 2003a, 2003b, 2004, 2005; Knobe & Mendlow, 2004; Nadelhoffer, 2004a, 2004b, 2004c, 2006; McCann, 2005; Adams & Steadman, forthcoming; Machery, forthcoming; Nichols & Knobe, forthcoming; Nichols & Ulatowski, forthcoming; Phelan & Sarkissian, forthcoming)

Evaluative –also influence Intentionality judgments attributions

The HARM vignette

The VP 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 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.

Did the chairman harm the environment intentionally?

Participants’ dominant (70-80%) response was to say that the chairman harmed the environment intentionally.

(Knobe, 2003; see also Nichols & Ulatowski, forthcoming)

The HELP vignette

The VP 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.

Did the chairman help the environment intentionally?

Participants’ dominant (70-80%) response was to say that the chairman did not help the environment intentionally.

(Knobe, 2003; see also Nichols & Ulatowski, forthcoming)

This asymmetry in responses to cases like HARM and HELP has been coined the “Knobe effect”.

Those concerned with folk psychology must explain these findings.

Our plan:(1) articulate a new account of the Knobe effect;

g

(2) offer empirical support for this account;g

(3) respond to two objections to our account.

Explaining the Knobe effect

Asymmetrical responsibility

Intuitively, agents are typically to some extent blameworthy, criticizable, or otherwise negatively responsible when they engage in an action that they know will bring about a bad outcome (or, an outcome which they have reason to not bring about).o This is so even if that outcome is simply a side-effect of an

intended outcome.

But agents are not typically praiseworthy, laudable, or otherwise positively responsible merely for bringing about a good outcome (or, an outcome which they have reason to bring about).

In order to be to some extent positively responsible for bringing about a good outcome, intuitively one must bring about that outcome for the right reasons—that is, because one has reason to bring it about.

For example…

In HARM the chairman presumably knew that his action would have a bad outcome, thereby making blame seem warranted.

After all, he was given a reason not to implement the new program—namely, that it would harm the environment—yet he still implemented the program anyway.

On the other hand, in HELP the chairman brought about a good outcome—namely, helping the environment—but did not do so for the right reasons.

Since he implemented the program simply because he desired to make money, praise seems unwarranted.

Responsibility and Intentionality

Typically, those who intentionally act to bring about a bad outcome are negatively responsible and those who intentionally act to bring about a good outcome are positively responsible.

That is: typically,good/bad action + intentionality = positive/negative

responsibility

In this way, intentionality commonly connects the evaluative status of actions to the responsibility of actors.

We normally infer positive/negative responsibility from both the goodness/badness of an action and the presence/absence of intentionality.

The action was intentional.

The action was good/bad.

The actor was positively/ negatively

responsible.

But other inferences are possible.

Actors are typically held to be responsible for good/bad actions only if they act intentionally.

So, the responsibility of actors serves as an indicator for intentionality.

We can infer the value of n from the equation

m + n = o

given specific values for m and o.

Likewise, using the following “equation”

good/bad action + intentionality = positive/negative responsibility

it is possible to infer (defeasibly) the presence/absence of intentionality given specific information regarding the goodness/badness of an action and the responsibility of an actor.

Suppose, for instance, that one judges that the outcome of a given agent’s action A was bad, and that the agent knew this but A ed nevertheless.

So, one judges that the agent was blameworthy for having A-ed.

One might then reason:o the agent is blameworthy, and thus

responsible, for having A-ed;o typically, an agent who is responsible for

having A-ed, A-ed intentionally; so, probably the agent A-ed intentionally.

For example…

Viewing the HARM chairman as blameworthy for a bad outcome puts one in a position to infer that he acted intentionally.

Harming the environment was foreseen.

Harming the environment

was bad.

The chairperson was blameworthy.

The chairperson acted intentionally.

Viewing the HELP chairman as not praiseworthy not only puts one in a position to infer that he did not act intentionally, it actually creates a reason to not so infer.

For if the chairman helped the environment (a reputedly good action) intentionally, then there would be good reason to judge him praiseworthy.

Since one judges him to be not praiseworthy, one infers that he did not help the environment intentionally.

Helping the environment was foreseen.

Helping the environment was good.

The chairperson was not

praiseworthy.

The chairperson did not act

intentionally.

Helping the environment was not done for the right

reason.

Explaining the Knobe effect

Two factors:1. assessments of positive/negative responsibility are

asymmetrical;2. intentionality commonly connects the goodness/badness of

actions to the responsibility of actors, the latter of which alone typically implies intentionality.

Participants in the relevant studies typically blamed in the (reputedly bad) HARM scenario, but did not praise in the (reputedly good) HELP scenario.

Because one may attribute intentionality when responsibility (and goodness/badness) is present, this resulted in an attribution of intentionality in HARM but not in HELP.

Empirical support

122 participants

HARM HELP

In both cases, participants were significantly more likely (p < .05) to judge that the chairperson acted intentionally when they stated both that the action was good/bad and that the chairperson was praiseworthy/blameworthy than when they only agreed to one or neither of these.

BAD 92.6%

BLAMEWORTHY 88.4%

INTENTIONAL 64.8%

GOOD 9.9%

PRAISEWORTHY 14.9%

INTENTIONAL 4.1%

The central role: responsibility

In both cases, participants were significantly more likely to judge that the chairperson acted intentionally when they stated that the chairperson was praiseworthy/blameworthy than when they did not.

On the other hand, in both cases, participants were no more likely to judge that the chairperson acted intentionally when they stated that the action was good/bad than when they did not.

Partial correlations reveal that judgments of goodness/badness considered alone were actually negatively correlated with intentionality attributions. o This indicates that judgments of goodness/badness became relevant

only when coupled with assessments of praise/blame.

These findings strongly support our account (and simultaneously disconfirm other accounts – e.g., the badness account).

Because the chairperson in HARM but not HELP was judged to be responsible, the chairperson in HARM but not HELP was judged to have acted intentionally. Hence the asymmetry.

Two Objections

1. The DECREASE vignette Susan is the president of a major computer corporation. One day, her

assistant comes to her and says, “We are thinking of implementing a new program. If we actually do implement it, we will be increasing sales in Massachusetts but decreasing sales in New Jersey.” Susan thinks, “According to my calculations, the losses we sustain in New Jersey should be a little bit smaller than the gains we make in Massachusetts. I guess the best course of action would be to approve the program.”“All right,” she says. “Let’s implement the program. So we’ll be increasing sales in Massachusetts and decreasing sales in New Jersey.”

Preliminary results (N < 25): o few participants judged that decreasing sales was bad;o few participants blamed Susan for decreasing sales;o but many participants judged that Susan decreased sales intentionally.

Popular conclusion: blame cannot explain the Knobe effect. (Knobe & Mendlow, 2004; Knobe, 2006; Nichols & Ulatowski, forthcoming;

Phelan & Sarkissian, forthcoming)

1. The DECREASE vignette Susan is the president of a major computer corporation. One day, her

assistant comes to her and says, “We are thinking of implementing a new program. If we actually do implement it, we will be increasing sales in Massachusetts but decreasing sales in New Jersey.” Susan thinks, “According to my calculations, the losses we sustain in New Jersey should be a little bit smaller than the gains we make in Massachusetts. I guess the best course of action would be to approve the program.”“All right,” she says. “Let’s implement the program. So we’ll be increasing sales in Massachusetts and decreasing sales in New Jersey.”

Preliminary results (N < 25): o few participants judged that decreasing sales was bad;o few participants blamed Susan for decreasing sales;o but many participants judged that Susan decreased sales intentionally.

Popular conclusion: blame cannot explain the Knobe effect. (Knobe & Mendlow, 2004; Knobe, 2006; Nichols & Ulatowski, forthcoming;

Phelan & Sarkissian, forthcoming)

Responses

1. Our account does not offer a generally necessary condition for intentionality attributions. o Factors other than responsibility, such as explicitly deliberating about

whether to A and then A-ing, will often be sufficient for holding that A was done intentionally.

o Indeed, the source of participants’ intentionality attributions in DECREASE was presumably Susan’s explicit deliberation.

So, the fact that participants attributed intentionality to Susan’s action while failing to judge her blameworthy is entirely consistent with our account.

2. Since a similar vignette in which Susan increased sales would also presumably invite intentionality attributions, DECREASE wouldn’t elicit the Knobe effect.

In order to test the Knobe effect, we need scenarios that elicit the Knobe effect.

The DECREASE2 vignette

The VP of a company went to the chairperson of the board and said, “We are thinking of starting a new program. It will help us increase profits, but it will also decrease sales in New Jersey.” The chairperson of the board answered, “I don’t care at all about decreasing sales in New Jersey. I just want to make as much profit as I can. Let’s start the new program.” They started the new program. Sure enough, profits increased and sales in New Jersey decreased.

Results (N = 122):o mirror HARM: frequent intentionality attributionso assessments of responsibility were significantly correlated with

the frequent intentionality attributions

The INCREASE vignette

The VP of a company went to the chairperson of the board and said, “We are thinking of starting a new program. It will help us increase profits, and it will also increase sales in New Jersey.” The chairperson of the board answered, “I don’t care at all about increasing sales in New Jersey. I just want to make as much profit as I can. Let’s start the new program.” They started the new program. Sure enough, profits increased and sales in New Jersey increased.

Results (N = 122):o mirror HELP: infrequent intentionality attributionso assessments of responsibility were significantly correlated with

the infrequent intentionality attributions

Empirical support

These findings support our account: o a judgment of both a good/bad action and a

responsible actor typically generates an intentionality attribution—and, of the two, the responsibility of the actor plays the central role.

Consequently, rather than challenging our account, these cases actually support it.

2. A “pointless mechanism”?

It has been argued that: o intentionality must play a useful folk psychological role in

evaluative judgments,o but the view that assessments of responsibility influence

intentionality attributions makes the psychological representation of intentionality a “pointless mechanism”.

(Knobe & Mendlow, 2004)

But using the presence of responsibility to infer the presence of intentionality on some occasions clearly does not make the psychological representation of intentionality a pointless mechanism.

More generally, because of the complexity of folk psychology and indiscriminateness of inference, it seems implausible to deny (a priori) that assessments of responsibility may sometimes influence intentionality attributions.

Thanks to the following for comments and discussion:

Derek Ball Jonathan Dancy

Joshua KnobeMarc Moffett

Shaun Nichols Mark PhelanGeorge SherEd Sherline

Also, thanks to Christin Covello, Jerry Cullum, and Piper Grandjean for assistance with data collection/entry.