Newell & Shanks (N&S) establish the elements of a decision by using the lens model (Brunswik Reference Brunswik1952). According to their framework, decisions can be either conscious or unconscious: In the former, all five stages of the lens model are supervised by consciousness, whereas in the latter at least one of the five elements is unconsciously performed. In our opinion, this is an extremely sharp distinction that leaves out of the picture aspects of a crucial importance in action selection, such as habits or habitual decision making.

Human agents make many decisions every day, some of which are fully unnoticed. Considering the role of consciousness in the course of decision making, we propose three categories in which a decision can be included: (1) conscious decisions, (2) retrospective attributions to unconscious behaviour, and (3) non-conscious but controlled decisions. The first type refers to deliberative decisions, which are made when facing a problem that requires a high cognitive load, especially involving a novel situation. Retrospective attributions are actions performed under low or non-existent levels of consciousness, and whose meaning is attributed a posteriori according to its effect; they include the illusion of conscious will (Wegner Reference Wegner2004), whose extreme interpretation suggests that consciousness is just a result of brain function. In our opinion, N&S only consider these two categories throughout their review, skipping the third type: rationally motivated decisions that are unconsciously performed but continuously open to conscious control. This omission disregards the dynamic aspect of decision making, which is contributed by learning. To illustrate the importance the third category of our classification has in human decision making, we will focus on technical habits.

Technical habits such as driving, painting, playing an instrument, or handwriting involve a high number of “decisions” – understood as the selection of a cognitively motivated course of behaviour – in their performance. Considering the lens model, most of them are unconscious in every stage of the process, since the agent does not even realise that a decision has been made. However, when a difficulty or something novel appears, consciousness immediately regains control of the process. At this point, it becomes evident the importance of previous learning and the dynamic aspect of decision making. Through habit learning, the agent transfers some particular actions from a conscious to an unconscious performance, without losing the capacity to consciously intervene at any time. Neuroscientific literature often identifies habits with automatic behaviour, interpreting them as a simple stimulus–response pair (see Seger & Spiering [Reference Seger and Spiering2011] for a review). As it has been recently proposed (Bernacer & Gimenez-Amaya Reference Bernacer, Gimenez-Amaya, Suarez and Adams2012), we believe that this view considers only the unconscious aspect of habit's performance and ignores the continuous access that consciousness has to regulate the process. In fact, we propose that habitual decision making should be considered an overall conscious rather than an unconscious operation, because it is consciously initiated and accessible to conscious control at all times. Another reason to support habitual decision making as an overall conscious process can be found in its comparison with instincts (Brigandt Reference Brigandt2005; Lorenz Reference Lorenz, Danielli and Brown1950). From the point of view of ethology, both – learned habits and innate instincts – include unconscious stages that contribute to a final conscious and cognitive goal. However, there is an important distinctive feature between them: When an instinct is abruptly interrupted, the whole operation ends. By contrary, when a non-conscious but controlled process is suspended, it is substituted by the conscious performance of the action.

We acknowledge new difficulties arise when including habits in decision making. The first of them is the most basic: how to delimitate a decision in this scenario. When a pianist is improvising, is he or she deciding to play each note, each chord of the whole melody? In our opinion, as it happens with actions, a decision has to be defined by its aim (Murillo Reference Murillo, Sanguineti, Acerbi and Lombo2011). Thus, in this particular example, the pianist decides to improvise. However, denying that this conscious decision – improvising – initiates many other non-conscious decisions – each key stroke, which has become non-conscious through habit learning – is a partial view of the process. For that reason, we believe that a more flexible understanding of the role of consciousness in decision making may be beneficial to have a more accurate view of the whole process.

In addition, it should also be considered the experimental difficulties to assess habit learning in the laboratory. The acquisition of a habit is usually measured by the decrease in the number of errors and the reduction of serial reaction times when repeatedly performing a sequence of movements (Aznarez-Sanado et al. Reference Aznarez-Sanado, Fernandez-Seara, Loayza and Pastor2013). Furthermore, experiments usually include a second task to be performed at the same time as the “automatic behaviour” (Wu et al. Reference Wu, Kansaku and Hallet2004). This approach involves the additional difficulty of distinguishing which brain activity pattern corresponds to which task. Although the neural bases of consciousness is a highly debated matter, cognitive cortical and subcortical areas had been associated to conscious performance of actions, whereas motor and premotor regions of the cortex, the posterior striatum, and cerebellum are considered to be in charge of subconscious processes. A recent study assessed early motor learning in the course of a continuous motor task by using functional magnetic resonance imaging. Brain activation progressively decreased in prefrontal cognitive regions and, conversely, increased in motor-related brain areas. Interestingly, an enhanced connectivity between the posterior putamen – a motor-related region – and the hippocampus was found, which supports the hypothesis of interactive cortico-subcortical memory systems in the course of learning (Fernandez-Seara et al. Reference Fernandez-Seara, Aznarez-Sanado, Mengual, Loayza and Pastor2009). To the best of our knowledge, a reliable neuroscientific study of technical habits is yet to be achieved. For that reason, we believe it is extremely important to lay solid theoretical foundations for an adequate experimental approach.

To conclude, N&S give an interesting overview about the lack of reliable evidence to demonstrate the role of unconscious drivers in decision making. However, their static framework does not allow the inclusion of habit learning, a fundamental element in decision making that involves a continuous transference of actions between the conscious and unconscious domains of the mind.