There are many things to like about the target article, and we thank Newell & Shanks (N&S) for their efforts: They will definitely result in a better science, which is what we all want, after all. Our gratitude, however, should not be mistaken for agreement: We widely disagree. For lack of space we will focus on two issues, one theoretical and one empirical.

Imagine the Snamuh, a species with brains and cognition just like ours. Unlike us, however, the Snamuhs are very proud of their unconscious: They believe that it is what sets them apart from the rest of the animal kingdom. Consciousness, they think, does not play a significant role in their lives.

Recently, papers in Snamuh-land's scientific journals suggested that consciousness matters. Curious about these developments, two of Snamuhs' best cognitive scientists examined the evidence in light of strict criteria. Among them are the relevance criterion (“assessments should target only information relevant to the behavior,” N&S Table 1), and the sensitivity criterion (“assessment should be made under optimal retrieval conditions,” N&S Table 1).

The scientists were not convinced by the data. After all, conscious information might be “contaminated” (Wilson & Brekke Reference Wilson and Brekke1994) by non-conscious processes, rendering the task of meeting the relevance criterion difficult. Similarly, assessing sensitivity of conscious measures is not a trivial task, threatening the sensitivity criterion. Accordingly, they concluded “conscious influences – if they exist at all – have limited and narrow effects.”

The moral is straightforward: N&S's default is that decisions are conscious. This is their H0. The evidence to the contrary, they suggest, is not convincing enough (we disagree, of course). Hence, H0 is not rejected and consciousness it is. Yet, the Snamuhs used the same scientific logic. But with the opposite H0, their conclusion was diametrically opposed. So which one is it?

We propose that the default view is less likely than it might seem. First, our consciousness is notoriously limited in its capacity (e.g., Baddeley & Hitch Reference Baddeley, Hitch and Bower1974; Kahneman Reference Kahneman1973). Given the sheer volume of problem solving, goal pursuits, self-regulation, and the like, that we engage with, it seems reasonable to suspect that we benefit from non-conscious progress too. Second, our best theories (to date) suggest that the type of consciousness that we have is relatively new in the evolutionary sense. It seems unlikely that basic functions such as decision making necessitate this “new” invention (see Hassin Reference Hassin2013; Hassin & Sklar, in press; Sklar et al. Reference Sklar, Levy, Goldstein, Mandel, Maril and Hassin2012).

Given this assessment, it will be fruitful to conduct research with the Snamuhs' H0. We urge scientists who argue that consciousness has a role in a cognitive function F (whatever F might be) to adopt strict criteria and measurements, similar to those adopted by researchers of the unconscious. For example, have them measure all of the relevant unconscious effects and contents and devise measures that are unaffected by irrelevant unconscious data. To the best of our knowledge, there is little discussion of how one should go about addressing these issues (but see Jacoby Reference Jacoby, Lindsay and Toth1992) – nothing that is even reminiscent of the debates that allowed the science of the unconscious to improve so much over the decades (with the kind help of the authors of the target article).

In the realm of high-level cognition, a role of consciousness is usually intuitively assumed or inferred from (what N&S call) weak evidence for unconscious processes. We would very much like to see positive, strong evidence of the kind described above. We believe there must be. Until this kind of research blooms, however, we cannot really estimate to what extent conscious processes affect decisions.

Our second point has to do with the literature coverage. For lack of space, we will focus on the subliminal priming section.

We are happy to have been reviewed in this section; we are very proud of this work (see also Carter et al. Reference Carter, Ferguson and Hassin2011). But there is a lot of evidence that is not reviewed there. For example, thirsty participants who had been subliminally primed with thirst-related words drank more than non-primed participants (Strahan et al. Reference Strahan, Spencer and Zanna2002; see also Veltkamp et al. Reference Veltkamp, Custers and Aarts2011). Extending these findings, Karremans et al. (Reference Karremans, Stroebe and Claus2006) showed effects of subliminal brand names (see also Bermeitinger et al. Reference Bermeitinger, Goelz, Johr, Neumann, Ecker and Doerr2009; Verwijmeren et al. Reference Verwijmeren, Karremans, Stroebe and Wigboldus2011). Subliminal primes also affect choices in “free choice” paradigms (Kiesel et al. Reference Kiesel, Wagener, Kunde, Hoffmann, Fallgatter and Stöcker2006; Klapp & Haas Reference Klapp and Haas2005; Schlaghecken & Eimer Reference Schlaghecken and Eimer2004), and we recently demonstrated effects of subliminal priming on choice, in a task that used both objective and subjective measures on a trial-by-trial basis (Milyavsky et al. Reference Milyavsky, Hassin and Schul2012).

Related research examined processes that are likely to affect decisions and has documented subliminal/implicit effects on executive functions and working memory (Hassin et al. Reference Hassin, Bargh and Zimerman2009; Lau & Passingham Reference Lau and Passingham2007; van Gaal et al. Reference van Gaal, Ridderinkhof, Fahrenfort, Scholte and Lamme2008), affect (e.g., Monahan et al. Reference Monahan, Murphy and Zajonc2000), numbers and arithmetic (e.g., Opstal et al. Reference Opstal, Lange and Dehaene2011; Sklar et al. Reference Sklar, Levy, Goldstein, Mandel, Maril and Hassin2012), incentives (e.g., Bijleveld et al. Reference Bijleveld, Custers and Aarts2009), goal pursuit (e.g., Ferguson Reference Ferguson2008), information integration (Mudrik et al. Reference Mudrik, Breska, Lamy and Deouell2011), fear (Raio et al. Reference Raio, Carmel, Carrasco and Phelps2012) and fear of death (e.g., Arndt et al. Reference Arndt, Greenberg, Pyszczynski and Solomon1997), anchoring (e.g., Mussweiler & Englich Reference Mussweiler and Englich2005), self-evaluation (Mussweiler et al. Reference Mussweiler, Rüter and Epstude2004), and attitudes (e.g., Loersch et al. Reference Loersch, McCaslin and Petty2011), to name, really, just a few (see Sklar et al. Reference Sklar, Levy, Goldstein, Mandel, Maril and Hassin2012 for more).

The authors raise a few general worries about these kinds of studies. Yet, the debate about measuring awareness has been with us for decades. Yes, there are justified concerns, but also, yes, there are good answers, which are met by counterarguments, and replies, and the development of new measures, and so on and so forth: scientific dialectics at its best. The question of subliminal semantic processing, for example, has been with us for decades too, and recent reviews suggest that it is possible (Kouider & Dehaene Reference Kouider and Dehaene2007; Van den Bussche et al. Reference Van den Bussche, Van den Noortgate, Reynvoet, Bussche and Noortgate2009).

To conclude, we believe that the data for subliminal effects on various aspects of decisions is much larger, and much stronger, than the data reviewed here. We therefore believe that the conclusions of this section do not adequately represent the state of affairs in this blooming area of the cognitive sciences.