How research on persuasion can inform dual-process models of judgment

Abstract

De Neys makes some useful points regarding dual-process models, but his critique ignores highly relevant theories of judgment from the persuasion literature. These persuasion models predate and often circumvent many of the criticisms he makes of the dual-process approaches he covers. Furthermore, the persuasion models anticipated some of the correctives to dual-process models that he proposes.

De Neys aims to provide a broad critique of prevailing dual-process and system (DP/S) models of judgment in “key fields,” as well as introduce a more viable approach (see Petty & Briñol, 2008, on dual-process vs. system frameworks). However, his critique fails to consider theories from the persuasion literature such as the heuristic-systematic (HSM; Chaiken, Liberman, & Eagly, 1989) and elaboration likelihood (ELM; Petty & Cacioppo, 1986) models that are clearly relevant and more highly cited than several of the covered DP/S approaches. Critically, the relevant persuasion models often agree with and predate the core points De Neys makes, and have already addressed some of the key challenges he poses. De Neys emphasizes how his new model is superior to prevailing DP/S models, but ironically his new model is better largely because it mimics features of the earlier persuasion models that were ignored. We illustrate our points largely using the ELM because we are intimately familiar with it, but also because there are numerous ELM studies that support our points (Petty & Briñol, 2012).

The first critique De Neys's offers of DP/S models is that they rely on exclusivity – the notion that fast (relatively low thought) and slow (relatively high thought) systems should yield different judgments. In contrast, the author proposes that high- and low-thought processes can: (1) “cue the same response” and (2) might not have “the same features.” These two ideas are fundamental to the ELM which explains how and why high- and low-thought processes can result in the same outcome under some circumstances but different outcomes under others. For example, is it better for persuasion to give people 3 or 9 message arguments? The ELM holds that it depends on whether the arguments are cogent or specious and whether people are engaged in relatively high or low amounts of thinking. When thinking is high, people evaluate the merits of the arguments, but when thinking is low, they are more likely to rely on simply counting the arguments using the heuristic – the more the better. Thus, when the arguments are strong, 9 arguments produce more persuasion than 3 regardless of the amount of thinking because processing for merit and counting produce the same outcome. However, when the arguments are weak, the high- and low-thought processes lead to different outcomes. Under low thinking, 9 weak arguments are still more persuasive than 3 because of the quantity heuristic. But, under high thinking, 9 weak arguments are less persuasive than 3 because they produce more negative thoughts (Petty & Cacioppo, 1984).

Regarding the second point, the ELM explains that even though the persuasion outcome is the same under high and low thinking when the arguments are strong (i.e., 9 > 3), the “features” of enhanced persuasion under 9 arguments can differ because the processes that led to that superiority are different. Specifically, the evaluations induced by 9 arguments over 3 under high thinking are more likely to persist over time, resist change, and guide behavior than the very same evaluations induced via a lower thought heuristic process (Haugtvedt & Petty, 1992). Thus, although we agree with the author's insight, this notion has been evident in the ELM for a long time (for parallels in ELM-guided work on numerical anchoring, see Blankenship, Wegener, Petty, Detweiler-Bedell, & Macy, 2008).

Another critique of DP/S models De Neys offers is that they do not explain how and when people might switch between low- and high-deliberation modes. De Neys also postulates that people always switch from low to high deliberation. In contrast, the ELM holds that people can start their processing at high elaboration. For instance, when a person initially views a particular judgment as important enough to think about carefully, there is no need to start with or generate a low-deliberation response first that then has to be corrected (Petty & Cacioppo, 1990). That is, low deliberation is not assumed to be the default mode. Rather, many variables determine whether an initial judgment results from high or low deliberation or whether people shift from one mode to another (Carpenter, 2015).

To explain when people move from a low- to a high-deliberation mode, De Neys proposes that it stems from uncertainty about the correct output (i.e., when low- and high-deliberation modes produce different outcomes). When uncertainty reaches a particular threshold, people shift to high thinking and this deliberation ceases when uncertainty drops below that threshold. Although De Neys's certainty threshold notion is quite reasonable, we note that it parallels the earlier sufficiency principle from the HSM (Chaiken et al., 1989). Furthermore, according to the persuasion models, in addition to uncertainty (e.g., stemming from ambivalence; Petty, Briñol, & Johnson, 2012), many other variables have been shown to motivate and/or enable enhanced deliberation (e.g., personal relevance of the judgment, responsibility for the judgmental outcome, etc.; Petty & Wegener, 1998).

Another critique is that DP/S models largely hold that the flawed (biased) outcomes occur when the output of low thinking is not corrected by high thinking. In contrast, De Neys proposes that deliberation “does not magically imply that the resulting response will be correct” (target article, sect. 3.3, para. 2). Yet again, persuasion models had already proposed that the amount of thinking and the extent of bias in that thinking are orthogonal (Petty & Cacioppo, 1990; for an example about stereotyping, see Wegener, Clark, & Petty, 2006). Thus, high thinking can sometimes lead to an even more flawed (biased) judgment than low thinking when, for example, a prime biases an initial (fast) judgment that then guides and contaminates the subsequent thinking (Petty, 2001).

In sum, although De Neys makes some reasonable points, a number of those points parallel principles previously proposed and documented in research examining relevant persuasion theories. By ignoring those frameworks, including their applications beyond the persuasion context (e.g., Petty & Briñol, 2014) and especially in judgment and decision-making domains where the criticized DP/S approaches have dominated (e.g., see Wegener, Petty, Blankenship, & Detweiler-Bedell, 2010), De Neys missed an opportunity to provide a more complete and integrative critique of DP/S models.