- If the slides are not working, or you prefer them full screen, please try this link
Greene offers an elaborate dual-process theory of ethical cognition, one which incorporates controversial claims about consequentialism and emotion.1 As these claims are neither essential features of a dual-process theory nor necessary for the overall argument we are developing, we may consider a stripped-down dual process theory instead.
According to this theory:
Two (or more) ethical processes are distinct in this sense: the conditions which influence whether they occur, and which outputs they generate, do not completely overlap.
One process is faster than another: it makes fewer demands on scarce cognitive resources such as attention, inhibitory control and working memory.
A key feature of the stripped-down two-systems theory is its theoretical modesty: it involves minimal commitments concerning the particular characteristics of the processes. Identifying characteristics of the process is a matter of discovery.
To make use of existing evidence, we have to add an auxiliary assumption to the dual-process theory:
Only the slow process ever flexibly and rapidly takes into account differences in the more distal outcomes of an action.
Prediction 1: Increasing cognitive load will selectively slow characteristically consequentialist responses. This prediction has been confirmed (Greene, Morelli, Lowenberg, Nystrom, & Cohen, 2008).
Prediction 2: Limiting the time available to make a decision will reduce characteristically consequentialist responses. This prediction also appears to have been confirmed:
‘The model detected a significant effect of time pressure, p = .03 (see Table 1), suggesting that the slope of utilitarian responses was steeper for participants under time pressure. […] participants under time pressure gave less utilitarian responses than control participants to scenarios featuring low kill–save ratios, but reached the same rates of utilitarian responses for the highest kill–save ratios’ (Trémolière & Bonnefon, 2014, p. 927).2
On the face of it, then, the dual-process theory appears well supported by evidence (and Greene, 2014 cites much further evidence). We may therefore accept it for now.
We will consider some more evidence for, and against, the dual-process theory.
Can be found in the notes to another course:
What is the strongest evidence in favour of the stripped-down two-systems theory of ethical cognition? Greene (2014) cites many studies. Here we will consider three of them:
Suter & Hertwig (2011) is an example of a relatively simply study which provides evidence in favour of the dual process theory plus auxiliary hypothesis.
One limit of this study is that it does not involve any variation in the distal outcomes of actions. This is relevant because the auxiliary hypothesis is about how different processes are differently influenced by distal outcomes.
Although not designed with exactly this in mind, Trémolière & Bonnefon (2014) does observe responses to otherwise similar actions with different distal outcomes. However, the findings are not predicted by the dual process theory and auxiliary hypothesis.4
One limit of both Suter & Hertwig (2011) and Trémolière & Bonnefon (2014) is that they treat responses as either consequentialist or not. These studies are sometimes presented as comparing consequentialist with deontological responses; but this cannot be accurate because failing to respond as a consequentialist would does not make you a deontologist (you may be neither).
Conway & Gawronski (2013) overcome this limit in addition to observing responses to otherwise similar actions with different distal outcomes. It is one of the strongest tests of the stripped-down dual process theory and its auxiliary hypothesis. These authors find, as predicted, that higher cognitive load reduces sensitivity to outcomes while not affecting sensitivity to moral prohibitions (such as on killing).
Conway & Gawronski (2013) are also important because they introduce process dissociation in moral psychology. Although difficult to understand, this is a powerful method for testing two-systems theories.5
Recall that Suter & Hertwig (2011) provide evidence that time pressure makes participants less sensitive to distal outcomes. Bago & Neys (2019) consider what happens when subjects first make a moral judgement under time pressure and extraneous cognitive load and then, just after, make another moral judgement (in answer to the same question) with no time pressure and no extraneous cognitive load. They report:
‘Our critical finding is that although there were some instances in which deliberate correction occurred, these were the exception rather than the rule. Across the studies, results consistently showed that in the vast majority of cases in which people opt for a [consequentialist] response after deliberation, the [consequentialist] response is already given in the initial phase’ (Bago & Neys, 2019, p. 1794).
This is an obstacle to considering Suter & Hertwig (2011)’s study as evidence for our dual-process theory of moral judgement.
Gawronski, Armstrong, Conway, Friesdorf, & Hütter (2017) note that reduced sensitivity to more distal outcomes could be consequence of a general preference not to do anything when under time pressure. They therefore extend the process dissociation model to include a preference for no action.
Separating sensitivity to distal outcomes from preferences not to act changes the picture:
‘The only significant effect in these studies was a significant increase in participants’ general preference for inaction as a result of cognitive load. Cognitive load did not affect participants’ sensitivity to morally relevant consequences’ (Gawronski et al., 2017, p. 363).
‘cognitive load influences moral dilemma judgments by enhancing the omission bias, not by reducing sensitivity to consequences in a utilitarian sense’ (Gawronski et al., 2017, p. 363).
While we should be cautious about putting too much weight on this study, these results do reveal that we cannot take Conway & Gawronski (2013) as evidence in favour of our dual-process theory and auxiliary hypothesis.
The two studies which conflict with the evidence for our dual-process theory also appear to conflict with each other. If Gawronski et al. (2017) is right about cognitive load, the participants in Bago & Neys (2019)’s study should have appeared to be less ‘utilitarian’ (as they describe it) when under cognitive load. This is because avoiding action would lead one to make judgements that Bago and Neys classify as non-utilitarian.
This is a sign that there may be something wrong with the way the studies are constructed, perhaps because the dual-process theories they are targeting are not well specified (e.g. involve too many independent bets being made simultaneously).
We may not yet have found sufficient grounds to accept or reject the stripped-down two-systems of ethical cognition.
Since automaticity and cognitive efficiency are matters of degree, it is only strictly correct to identify some processes as faster than others.
The fast-slow distinction has been variously characterised in ways that do not entirely overlap (even individual author have offered differing characterisations at different times; e.g. Kahneman, 2013; Morewedge & Kahneman, 2010; Kahneman & Klein, 2009; Kahneman, 2002): as its advocates stress, it is a rough-and-ready tool rather than an element in a rigorous theory.
See Paxton & Greene, 2010 for a compact overview of Greene’s theory. The theory has been presented in a variety of different ways (see, for example, Cushman, Young, & Greene (2010) for an alternative presentation). ↩
Later we will consider an alternative interpretation of the same findings due to Gawronski, Conway, Armstrong, Friesdorf, & Hütter (2018, p. 1006). ↩
This study is associated with a second prediction, which the results appear to disconfirm: limiting the time available to make a decision will reduce sensitivity to outcomes. ↩
Although the Trémolière & Bonnefon (2014)’s findings may be interpreted as disconfirming a prediction (as Gawronski & Beer, 2017, p. 669 propose), it would be incautious to rely on post hoc reinterpretations of findings. ↩