(Originally published Oct 9, 2012)

I’ve just returned from the Buffalo Experimental Philosophy Conference.1 Mark Alfano and I presented excerpts from our paper on “Virtue and Vice in the Business Context” (currently under review), focusing on attributions of compassion and callousness in the business context and how it mediates consumer behavior. (Our co-author Paul Stey, Ph.D. candidate and stat master, was not able to attend.)

Our new research builds off the Knobe Effect (also known as the side-effect effect), one of the most well known and replicated results in experimental philosophy. You might already be familiar with this literature, but let’s review just in case.  Josh Knobe discovered an interesting phenomenon regarding the intentionality of side effects. In Knobe experiments, participants are given a story of a protagonist, who is presented with a plan of action that includes a side effect. An interlocutor then makes salient a norm, which the side effect will either conform to or violate. The protagonist then explicitly disavows interest in the side effect and orders the plan of action to be initiated. The outcome is exactly as predicted; the plan brings about the intended primary result and the side effect conforms to or violates the norm.

In the original version a chairman of the board helped or harmed the environment, but merely as a side effect. Quite interestingly, Knobe found that people were much more willing to say the side effect was intentional when it violated the norm than when it merely conformed to it. This result has raised a number of questions and spawned number of follow up studies. Over the last decade, researchers have found it is not just intentionality, but also belief, knowledge, memory, desire, favor, and advocacy are asymmetrically ascribed. Others have found that it need not be a moral norm; aesthetic, legal, conventional, and descriptive norms also trigger the effect.2

We decided to look at two issues that haven’t been explored yet. First, we wanted to know if the Knobe Effect would make any difference to attributions of character traits. Second, we thought it worth exploring the implications the Knobe Effect has for business ethics. Today I’ll focus on the first.

Using a 2×2, between-subjects design, we presented participants with a version of the following vignette:

The vice-president of a manufacturing company was talking with the CEO.  The vice-president said, “We are thinking of implementing a new policy.  If we implement the policy, it will increase profits for our corporation. It will also mean that we have to fire 10% of our employees, many of whom will increase our workforce by 10%, hiring many people who would have difficulty finding other work.” The CEO said, “I don’t care at all about the employees. I just want to make as much money as possible. Let’s implement the policy.

They implemented the policy. Just as the vice-president had predicted, profits increased, 10% of the employees were fired, and many of them were unable to find other work / the workforce was increased by 10%, and many of the new workers would have been unable to find other work.

This is a standard Knobe case.  Our hypothesis, however, was that both virtues and vices are based upon primary (or main) effects.  So, we also gave other participants one of the following:

The vice-president of a manufacturing company was talking with the CEO.  The vice-president said, “We are thinking of implementing a new policy.  If we implement the policy, it will increase profits for our corporation. It will also mean that we have to fire 10% of our employees, many of whom will increase our workforce by 10%, hiring many people who would have difficulty finding other work.”  The CEO said, “I’ve been looking for ways to fire some of our employees, and of course I always want to increase profits. Let’s implement the policy.”

They implemented the policy. Just as the vice-president had predicted, profits increased, 10% of the employees were fired, and many of them were unable to find other work / the workforce was increased by 10%, and many of the new workers would have been unable to find other work.

We then asked our participants3 on a seven-point Likert scale how intentional, callous, and compassionate they regarded the hiring/firing, as well how likely they were to avoid purchasing products from this company, even if they had to pay more. Let’s start with intentionality:

Ratings of Intentionality by condition

Figure 1

As you can see, they did not regard hiring as norm-conforming side effect to be intentional; otherwise, intentionality holds steady. While perhaps the expected result, it is reassuring to confirm that the Knobe Effect doesn’t extend to primary effects as well.  What of the CEO’s callousness or compassion?

Figure 2

Figure 3

Put together, these two graphs tell an interesting story. You might think that the CEO who hires workers either as a side effect might be thought as compassionate as when hiring was the primary effect. Or, a slightly weaker claim, that the CEO in the hire/side-effect case at least won’t be callous. But neither the stronger or weaker claim matches the folk view. There is no “neutral” case, where the CEO is neither callous or compassionate. At least of CEO’s, callousness appears to be the default, only able to be overridden by a morally good, second primary effect. Being regarded as virtuous takes something special.

Finally, we asked about participants’ willingness to no longer buy products from the CEO’s company, even if it costs more. This is significant, since, to our knowledge this is the first study to consider what implications the Knobe Effect has for consumer behavior. Here’s what we found:

Figure 4

Participants reported being more willing to avoid purchasing when the CEO fires employees than when he hires employees. And within the fire conditions, they say they’re more inclined to avoid purchasing when the firing was a primary effect than a side effect. The really interesting result is that we found that the inclination to avoid purchasing from the company was highly correlated with participants’ ratings of the CEO’s callousness, r = .50, p < .001. This means that a participant’s rating of the callousness of the CEO was a predictor of how inclined they were to avoid purchasing from the CEO, F(1, 77) = 9.00, p < .01, ηp2 = .11.


1 Thank you to James Beebe and Paul Poenicke for organizing the conference. And thank you to the other presenters for their papers, as well as their comments and feedback to our presentation.

2 Alfano and I (with James Beebe) have previously presented a unifying account of the Knobe Effects.

3 N = 81, Mage = 34.4, 40 women