Newell & Shanks (N&S) argue that priming studies, although promising, are flawed by inadequate self-report measures of awareness. However, functional magnetic resonance imaging (fMRI) subliminal priming studies (which are perhaps more objective than self-report measures) should also be taken into account. In such work, differences in brain activation are observed between stimuli that are consciously versus subliminally perceived. Thus, in a recent meta-analysis by Brooks et al. (Reference Brooks, Savov, Allzen, Benedict, Fredriksson and Schioth2012) of subliminal fMRI studies, it is notable that non-conscious salient stimuli robustly activate the amygdala and hippocampus – brain regions linked to emotional arousal (Costafreda et al. Reference Costafreda, Brammer, David and Fu2008) and memory (Garcia-Lazaro et al. Reference Garcia-Lazaro, Ramirez-Carmona, Lara-Romero and Roldan-Valadez2012). In the meta-analysis, most studies used highly salient subliminal emotional faces, but subjects reported a lack of awareness of these faces, and there was no activation of the fusiform gyrus, a region of the occipital cortex well-known to be involved in conscious recognition of faces (Weiner & Grill-Spector Reference Weiner and Grill-Spector2012). Furthermore, when comparing brain responses to the same affective stimulus, presented either with or without awareness, our data indicate that there is a switch from insular cortex (associated with conscious interoceptive awareness) to posterior cingulate cortex activation (associated with default mode network and auto-affective processing), and that the anterior cingulate cortex (ACC) activates in response to both levels of perception (Meneguzzo et al., under review). The ACC is implicated in decision making, particularly with respect to conflict monitoring and error detection, as well as cognitive control (Carter & van Veen Reference Carter and van Veen2007), perhaps especially during high-anxiety states (Ursu et al. Reference Ursu, Stenger, Shear, Jones and Carter2003). Thus, decision-making processes may be mediated by areas such as the ACC, and they are influenced by non-conscious affective states (Panksepp Reference Panksepp2005).

Subliminal fMRI findings are consistent with a view that automatic activation of affective processes probably guides evolutionarily advantageous decision-making mechanisms (Panksepp Reference Panksepp2005; Reference Panksepp2011a). Such affective processes have evolved to respond efficiently to environmental threats and appetitive stimuli, are mediated by a range of neuroendocrine and neurotransmitter systems, and operate at least partially outside of awareness (Damasio Reference Damasio2010; Kahneman Reference Kahneman2011; Le Doux Reference Le Doux1996; Panksepp Reference Panksepp2005). The involvement of evolutionarily ancient mesolimbic brain regions, such as the amygdala, hippocampus, and striatum, as well as brain stem structures (e.g., the periaqueductal gray) during affective states, underlines their importance in adaptation (Alcaro & Panksepp Reference Alcaro and Panksepp2011; Panksepp Reference Panksepp2011b, Stein et al. Reference Stein, Solms and van Honk2006). The activation of these core brain regions presumably occurs with less conscious and less cortical processing in non-human species, and yet it is still meaningful in the decision-making process. Consistent with the findings emerging from the imaging of subliminal processes, neuroimaging work during exposure to threat cues highlights the role of similar brain regions. For example, Mobbs et al. (Reference Mobbs, Petrovic, Marchant, Hassabis, Weiskopf, Seymour, Dolan and Frith2007) demonstrated that the imminence of threat shifts brain activation from prefrontal regions to periaqueductal gray, particularly in terms of the degree of dread and lack of confidence in the decision to escape, supporting the notion that automatic brain processes are more prominent under highly arousing affective states.

Most subliminal fMRI studies convincingly assess the lack of awareness with a “forced choice” task post hoc (choosing between a novel and a previously subliminal image), which is further corroborated by the participants' inability to recollect verbally any subliminal affective stimulation during the scan (e.g., Brooks et al. Reference Brooks, Savov, Allzen, Benedict, Fredriksson and Schioth2012). Future fMRI studies using subliminal paradigms must ensure, as N&S emphasize, awareness checks that enable reliable self-report, are relevant to the task, immediately precede the subliminal stimulus, and are sensitive to the subliminal cues. Subliminal fMRI studies may provide an attractive methodology for overcoming some of the discrepancies outlined by N&S in the field of subliminal priming research. In line with N&S's arguments, given that subliminal fMRI studies have not yet probed decision making, this may be a particularly useful component of such work. An evolutionary theoretic framework may be useful in future work to provide an ecologically valid approach to choosing and analyzing relevant cues. Connectivity analyses in subliminal fMRI studies may be useful in probing not only localized brain function, but also in developing network models of the neural circuitry of non-conscious processes. Finally, rigorous studies of different subliminal cues may well be informative in the study of a spectrum of psychopathologies (Stein Reference Stein1997).