We can begin with a question that goes unasked in De Neys's otherwise wide-ranging and insightful essay: Why is it that switching from system 1 to system 2 cognition is so often – though not always – an affectively loaded experience? Here I take the affective dimension of switching to be familiar. It appears, for instance, as the feeling of unease that one experiences when an initially routine and mindless task suddenly becomes more difficult and cognitively demanding. But notice that a similar feeling is typically absent when, for instance, cognitive resources are ramped up more gradually. So why do we see this difference? This commentary will not only draw attention to this neglected aspect of “system switching,” but also show how research in philosophy and the social and cognitive sciences sheds light on it. The result will be greater clarity on some of the open questions that De Neys raises at the end of his paper.

Like De Neys, philosophers have recognized that dual-process theorists owe us an explanation of system switching. But these philosophers also add substance by highlighting a distinctive family of metacognitive emotions as underlying many of these system 1 to system 2 transitions. These emotions are metacognitive in the sense that they function to regulate one's first-order cognitive processing; they are emotions in the sense that they are automatically engaged, motivationally laden feelings. More specifically, on these philosophical accounts, metacognitive emotions are viewed as system 1 forms of cognition that use heuristics to map occasions of positive/negative value to distinctive feelings, and then use these feelings to generate an “affective alarm” – a warning that (re)directs attention and engages a distinctive suite of system 2 processing (Arango-Muñoz, Reference Arango-Muñoz2011; de Sousa, Reference de Sousa, Brun, Doğuoğlu and Kuenzle2008; Kurth, Reference Kurth and Timmons2015, Reference Kurth2018a). So, returning to the earlier example, those feelings of unease are, on this metacognitive account, to be understood as automatically engaged responses to problematic changes in one's circumstances: Responses that function to warn of potential trouble and prompt higher cognitive processing to help one address the issue at hand.

Importantly, these theoretical proposals in philosophy are supported by empirical work from the social and cognitive sciences. For instance, we have research highlighting the role that feelings of familiarity play in the tip-of-the-tongue phenomenon (e.g., Schwartz & Metcalfe, Reference Schwartz, Metcalfe, Schwartz and Brown2014). This work identifies these feelings as mechanisms that function to engage and sustain conscious mnemonic effort after heuristic-based monitoring systems have identified an instance of failed memory with partial recall (e.g., occasions where you cannot remember a person's name, but sense that it begins with a “P”). Similar findings point to the role that anxiety plays in prompting deliberation in the face of difficult moral and political policy issues like affirmative action, immigration, and climate change (Fernando et al., Reference Fernando, Sivakumaran and Suganthi2016; MacKuen et al., Reference MacKuen2010; Valentino et al., Reference Valentino, Hutchings, Banks and Davis2008). Here these feelings of worry are seen as working to flag decisions like these as being particularly difficult, thereby engaging reflection and information gathering efforts to help one work through the issue.

Crucially for our purposes, in both cases the empirical findings implicate the metacognitive role of emotion in these system 1 to system 2 transitions: We see affect acting as a system 1 alarm – one that both shifts our attention to the problem at hand and engages system 2 resources to help us address it. Moreover, notice as well that these two examples characterize their target phenomena as involving distinctive triggers, felt experiences, and forms of higher cognitive engagement – features that suggest we have a family of distinct metacognitive emotions here, and not just a single, all-purpose mechanism (Arango-Muñoz, Reference Arango-Muñoz2011; Kurth, Reference Kurth2018b; Thompson, Reference Thomson, Evans and Frankish2009).

Here's why all this matters. First, the above examples provide us with concrete, empirically grounded models of how central instances of system 1 to system 2 switching may operate, thus responding to De Neys's call for “further fleshed out, fine-tuned, and developed” accounts these processes (target article, sect. 4, para. 1). Second, by understanding these switching mechanisms as emotions, these models shed new light on some of the “deliberation issues” that De Nays flags in section 4.3. For instance, given the familiar role that emotions play in directing attention and sustaining effort, we get the makings of explanations for questions about, respectively, the prioritization of deliberative effort across tasks and the amount of effort expended on a given occasion. Finally, if emotions have the alarm function suggested above, then we also get an answer to the question we started with: The reason why only some cases of system 1 to system 2 switching are affect-engaging is that only some of them are ones that our emotions have deemed to be alarm-worthy.