How Bad Was Abu Ghraib? It Depends on the Comparison

I have to admit that I've been avoiding the "framing science" discussion that's been going on in the science blogosphere recently, mostly because I'd rather talk about what framing is and how it works than two author's rather vague ideas about how to use framing in a particular area of discourse. And because the Science article has made framing a hot topic again, and because it is clear from much of the discussion that many are still very confused about what framing is (if I see someone describe framing as "spin," again, I'm going to throw something at them), I think it's important to talk about framing itself. As I was sitting around thinking about exactly how I wanted to approach the topic, Science Direct was kind enough to send me a social psychology topic alert that included a paper in press at the Journal of Experimental Social Psychology that highly some of the issues I think are important. The paper by Markman et al.1, and titled "'It Would Have Been Worse Under Saddam:' Implications of Counterfactual Thinking for Beliefs Regarding the Ethical Treatment of Prisoners of War," isn't actually about framing. It is, as the title suggests, about counterfactual thinking. Still, I think it illustrates the importance of the interconnection of language and representation for framing. I'll say more about that at the end of the post, but for now, let's talk about the paper.

First, a little background about counterfactual thinking. Counterfactuals generally come in two flavors: upward counterfactuals and downward counterfactuals. Most of the time when people think counterfactually, they use upward counterfactuals. These involve counterfactual situations that are better than the factual ones. For example, someone might think, "If I hadn't stayed up all night partying the night before, I wouldn't have failed my exam." Presumably, not failing is a better outcome than failing, so this is an upward counterfactual. Upward counterfactuals are the most common sort because we tend to reason counterfactually about negative events more than positive ones, and when we do so, we're more likely to think about how things could have been better than how they could have been worse. This helps us to understand why things went wrong, and since we tend to think counterfactually only when we could have done something differently, upward counterfactuals often lead to negative emotions like guilt and regret. Downward counterfactuals are less common than their upward cousins, mostly because we don't tend to second guess positive outcomes, and when something bad happens, we're less likely to think about how it could have been worse than how it could have been better. Still, they do occur. For example, someone might think, "If I hadn't stayed up all night studying, I wouldn't have gotten an A on the exam." Downward counterfactuals tend to produce positive emotions, like a sense of achievement, because they generally highlight a person's role in producing a positive outcome.

Markman et al. hypothesize that in addition to the different emotions produced by the two kinds of counterfactuals, they may also "shift the standard by which future outcomes are evaluated" (p. 4 of the manuscript). That is, upward counterfactuals may cause us to expect more positive outcomes, and to evaluate future outcomes accordingly. If someone thinks, "If I hadn't partied the night before, I would have gotten an A instead of a C on the exam," she might then expect A's on future exams, because the C was due to a one-time mistake. Downward counteractuals, then, should shift standards downwards, so that if someone thinks, "If I hadn't stayed up all night studying, I would have gotten a C instead of an A," she will consider herself a C student who managed to get an A once, and adjust her future expectations accordingly.

OK, that's enough about counterfactuals in general. Now to the research. Markman et al. were inspired by statements made by government officials and journalists/pundits in the wake of the Abu Ghraib prison abuse scandal to the effect that "It would have been worse under Saddam." Such statements are clear examples of downward counterfactuals: the factual outcome (abuse of prisoners by American personnel) is being compared to much worse treatment of prisoners by Saddam's regime. Given what we know about downward counterfactuals -- that they produce positive emotions and a downward shift in standards -- Markman et al. predict that such comparisons will cause people to feel less negatively about the Abu Ghraib scandal, and to shift their expectations for the future treatment of prisoners by Americans downward.

To test these predictions, they conducted a study in which they divided participants into three conditions. In the downward condition, participants first read a paragraph describing the Abu Ghraib abuses by American personnel, and then read a paragraph about Saddam's torture and execution of prisoners, and were asked to"make an argument that being at Abu Ghraib under Saddam's control would be worse than being there under U.S. control" (p. 6). In the upward condition, participants read the same paragraph describing the Abu Ghraib abuses, and then read a paragraph "that described the ethical treatment of prisoners by a small contingent of Danish soliers in a military prison based in the city of Al Quma" (p. 6), after which they were asked to ",make an argument that the ethical standards employed by the Danish in their treatment of Iraqi prisoners were better than the standards employed by the U.S. in treating Iraqi prisoners" (p. 6). Finally, in the control condition, participants only read the paragraph about the Abu Ghraib abuses.

Participants in all conditions were given two statements and asked to indicate how much they agreed with them on a 1 to 9 scale. The statements were, "I am morally outraged by the events that took place at Abu Ghraib," and "I feel good about what happened at Abu Ghraib." After rating their agreement with these statements, they were asked to write explanations for their level of agreement with the first statement. Finally, they answered four questions designed to assess their standards for the future treatment of prisoners by American personnel.

Markman et al. predicted that, consistent with the finding that downward counterfactuals produce more positive emotions, and upward counterfactuals more negative ones, participants in the downward condition would express more positive feelings (including less moral outrage) about the abuse scandal than participants in the upward or control conditions. They also predicted that participants in the downward condition would shift their standards downward, resulting in lower standards for the future treatment of prisoners under American control. And the data supported both hypotheses. First, participants in the downward condition had significantly more positive feelings about Abu Ghraib (on average, a score of 3.38 on a combination of the moral outrage and "I feel good" statements, with 9 being the most positive) than participants in the upward (an average of 1.85) and control (2.5) conditions. Participants in the downward condition also expressed less moral outrage in their explanations than did participants in the other two conditions. Second, participants in the downward condition had significantly lower standards of treatment (an average of 3.44, with 7 indicating the highest standards) than participants in the upward (5.6) and control (4.34) conditions. In this case, the difference between the upward and control conditions was also significant (it wasn't for the "positive feelings" measures, likely because the control participants' scores were so low).

At this point, it's probably clear why I think this study relates to framing. But in case it's not, consider two definitions from the framing literature, the first of frames and the second of the act of framing:

Frames are principles of selection, emphasis and presentation composed of little tacit theories about what exists, what happens, and what matters.2


To frame is to select some aspects of a perceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation.3

Quite clearly what the upward and downward counterfactuals in the Markman et al. study were doing was "Select[ing] some aspects of a perceived reality and mak[ing] them more salient," and they did so by using the ways in which people reason counterfactually, and how such reasoning affects our standards and beliefs. What government officials and journalists were doing by making statements about how much worse things would have been at Abu Ghraib if Saddam had still been in power was causing a frame shift from (presumably) knowledge about how western democracies (should) treat prisoners to knowledge about how Saddam treated prisoners. If done effectively, this can cause people to see the American abuses in a much more positive light, because they really weren't as bad as those that occurred under Saddam, and in addition, as the Markman et al. study shows, cause people to shift their standards for the future treatment of prisoners. In other words, people now had a different frame for thinking about American treatment of prisoners.

The lesson to take home from this is that language, or the act of framing in public discourse, and the cognitive processes that those acts utilize, are inseparable. I think this is an important lesson in part because, with a few exceptions, discussions of framing tend to focus more on the act of framing than on the frames (or representations) in people's minds, and the processes that operate on those frames (like those that produce the effects of counterfactual thinking). Take, for example, this comment by Matt Nisbet, one of the authors of the offending Science paper, in response to my last post on the topic:

However, I was struck by one point that fellow ScienceBlogger Matt Nisbet, one of the authors of the Science article that sparked this whole mess, made in comments to my post on the discussion. He wrote (emoticon removed, for your sanity):

In part what we have across the various disciplines studying framing is a classic "levels of analysis" problem. Some working at the micro and cognitive level, others working at the macro and sociological level.

What I hope the preceding discussion shows is that it's not, in fact, a "level of analysis" problem. Instead, it's a problem of people focusing on one of the deeply intertwined components of framing to the exclusion of others. Usually, the components excluded are the cognitive ones. But if you don't understand the cognitive ones, you won't be able to accurately prescribe effective acts of framing.

1Markman, K.D., Mizoguchi, N., & Mcmullen, M.N. (In Press). "It would have been worse under Saddam:" Implication of counterfactual thinking for beliefs reagarding the ethical treatment of prisoners of war. Journal of Experimental Social Psychology.
2Gitlin, T. (1980). The Whole World Is Watching: Mass Media in the Making and Unmaking of the New Left. Berkeley: University of California Press.
3Entman, Robert M. 1993. Framing: Toward clarification of a fractured paradigm." Journal of Communication,43(4), 51-58.

More like this

I've been meaning to post about this set of studies for a while, but because it's relevant to Chapter 4 of Lakoff's The Political Mind, I figured I'd better get around to it before I write the review of that chapter. It's been a while, but in the past, I've talked a lot about new theories of moral…
That's it! I'm never reading another imaging paper again, ever. OK, I might read one or two, and I might even post about them, but for now I'm telling myself, for my own sanity, that I'm never, ever, under any circumstances, going to read another imaging study. If you read my last post, or have…
It's been obvious to everyone who's written about politics since Aristotle that emotion plays an important role in political rhetoric and political judgment. With an increased focus on emotion in cognitive scientists, there has been a flourish of empirical work attempting to elucidate this role. I…
I've been posting about moral cognition anytime a new and interesting result pops up for a while now, and every time I think I've said before, though it bears repeating, that every time I read another article on moral cognition, I'm more confused than I was before reading it. Part of the problem, I…

I see that framing is not spin. But isn't framing an oft-used tool in spin?

Framing is an educational tool. How you use the tool is what's important. If you use a framing device to help your audience understand something better, which is what I think the framing Science article is about, that's good because it helps people understand. When you use framing to help people see things your way, that's spin.

How's this for a downward counterfactual? Our drug didn't increase the death rate! [But it was no better than placebo.] Buy buy buy!

Trial of Amgen Drug May Dispel Some Safety Concerns


"In an announcement that could dispel at least some of the recent safety concerns over a widely used class of drugs, Amgen said yesterday that its anemia drug Aranesp did not increase the death rate of lung cancer patients in a closely watched clinical trial.

In the trial, which involved 600 patients with small-cell lung cancer who were receiving chemotherapy, there was essentially no difference in survival between those who received Aranesp and those who received a placebo.

Amgen's stock surged $2.31, or 4 percent, to close at $62.32."

A personal favorite: GWB says we keep improving our carbon emission intensity. That is, emissions pre GDP dollar or per capita decline, so let's trumpet that in spite of the fact that carbon emissions in absolute terms are growing apace.

So the art of spin is to take an event's factual outcome and either:
amplify the good/bad effect of the event to promote a political agenda
minimize the good/bad effect of the event to protect a political agenda

Amplifying spin doctors use upward counterfactuals to exaggerate the effect.
Minimizing spin doctors use downward counterfactuals to downplay the effect.

Framing is used to make sure that the audience only hears counterfactuals that support the spin doctor's


If the audience is given an opportunity to hear from both amplifying and minimizing spin doctors on a

particular issue the effects of spin are cancelled out.

In the Markman et al. study it would have been interesting to have a group of participants read both the upward and downward counterfactual paragraphs and compare their reaction to the control condition.