Huang & Bargh (H&B) present an impressive review of research on unconscious sources of cognition and action. From our perspective, however, in their resolve to clear the path for the “selfish goal” metaphor, they may have missed an opportunity to make an even stronger case for the role that unconscious motivational processes play in cognition and action. Here, we outline such a framework (Eitam & Higgins Reference Eitam and Higgins2010; Eitam et al. Reference Eitam, Miele, Higgins and Carlston2013) and discuss its implications for the selfish goal metaphor.

The framework, relevance of a representation (ROAR), adds a second, motivational layer – which we called relevance – to one of the basic pieces of conceptual machinery in cognitive and social cognitive psychology: mental representations and their dynamics. The principal tenet of ROAR is that the activation of mental representations (externally or internally stimulated), and hence the accessibility of the concepts they represent, will be a function of the motivational relevance of those concepts.

In line with Higgins' (2012) analysis of motivation, within ROAR we (Eitam & Higgins Reference Eitam and Higgins2010) named this “goal” source of motivational relevance, that is, obtaining desired results, “value relevance.” As H&B convincingly demonstrate, this source of relevance is indeed pervasive. ROAR nicely accommodates the effects of goal activation described by H&B by specifying, for example, how so-called positive tags of an action representation (Custers & Aarts Reference Custers and Aarts2005) could lead to action selection. ROAR goes further to explain how, although many goal relevant objects exist in the environment, only a few (relevant) goals are activated (see Ferguson Reference Ferguson2007) because goals (and their activation) are themselves subject to the same relevance computation as any other knowledge representation.

But pursuing valued end-states, that is, goals, does not exhaust what motivates humans or other animals. Following Higgins (Reference Higgins2012), we specified two additional sources of motivational relevance – truth relevance (knowing what is real, what is correct) and control relevance (managing what happens), each of which affects activation independent of value relevance.

Truth relevance implements the mind's need to know what is really “out there.” If the mind is expecting something (see Bruner's Reference Bruner1957 “perceptual readiness”), once a sufficiently similar signal appears, the corresponding representation will be more strongly activated because of its high truth relevance. Through its effects on mental activation, truth relevance also impacts memory. This has been helpful in explaining the effects of “shared reality” (e.g., Echterhoff et al. Reference Echterhoff, Higgins and Groll2005). In these studies, participants are asked to read evaluatively ambiguous information about a target person and then present that person to an audience that likes or dislikes that person. Multiple studies have established that, although asked to reproduce the original (ambiguous) description, recall is biased towards the “shared reality” audience-tuned message. When participants are told later that the audience could not (vs. could) identify the target on the basis of the description they supplied, this “saying is believing” memory effect disappears. ROAR accounts for this disappearance in the failed communication condition in terms of reducing the truth relevance, and thus the accessibility, of the previously shared, and now “un-shared,” message.

The third source of relevance we proposed was relevance from control. This source of relevance implements the mind's necessity to know what can be successfully done/affected/controlled by the organism (or by other agents) in the environment. It enables integrating dissimilar results. For example, dopaminergic cells' phasic response is thought to be influenced by the degree of control the stimulus is associated with (Redgrave & Gurney Reference Redgrave and Gurney2006); task (i.e., goal) irrelevant stimuli that are contingent on one's action are nonetheless registered in humans (Band et al. Reference Band, van Steenbergen, Ridderinkhof, Falkenstein and Hommel2009); and action contingent effects are registered in babies prior to any association with goal-directed control (Verschoor et al. Reference Verschoor, Spapé, Biro and Hommel2013).

One could argue that the concept of a goal, selfish or not, could assimilate both the truth and control sources of relevance as being merely highly abstract goals. But this would make the notion of goal representation rather useless as an explanatory construct because it would always be possible to impute an even more abstract goal (such as “be part of God's universe”). Moreover, for any valued goal, there are still motivational questions regarding how real or imaginary that goal is (truth) or how its pursuit can be managed (control), and thus the need for distinguishing value relevance from truth and control relevance remains (see Higgins Reference Higgins2012; Silvestrini & Gendolla Reference Silvestrini and Gendolla2013).

We want to stress that in ROAR relevance is not a metaphor. Indeed, there are indications that a number of brain structures are involved in computing such relevance (for a review, see Eitam & Higgins Reference Eitam and Higgins2010). Importantly, the computation of relevance is affected by the organism's current state, including current needs and current knowledge. Thus, even if an activation goal may be selfish once activated, whether it is activated at all and how long it remains activated depends on its current and subsequent motivational relevance to the organism. In other words, per ROAR, the organism has “relevance control” over its goals as it does over other activated knowledge, whether it is aware of it or not.

We conclude with a note on stability (“consistency”) of behavior. We have documented the existence of representations that have cross-situational value relevance (faces; Eitam et al. Reference Eitam, Glass, Aviezer, Dienes and Higgins2013) and showed that these will operate regardless of a person's current task. Nonetheless, many representations have only transient relevance. Given that relevance determines activation, chronic relevance will lead to chronic activation and hence similar cognitions and actions across situations (consistency). Representations with transient relevance will make their appearance “inconsistently” or, rather, in a manner that is highly dependant on their current relevance to the situation at hand.

To summarize, although H&B highlight the central role of motivation in cognition and action, whether conscious or not, the chosen metaphor, that is, selfish goal, may unnecessarily limit their case in terms of capturing the scope of motivational effects on action and cognition and the degree of integration between cognitive and motivational influences. Unpacking their metaphor in terms of its underlying motivational-cognitive mechanisms could further strengthen its dominion.