There are two prominent notions of spontaneity discussed in the history of philosophy. The first corresponds with how we use the term in ordinary language: something that suddenly occurs, often without foresight and in a happenstance manner. The second is more idiosyncratic yet bears deeper roots: something that expresses our contributions to our behavior. This latter notion, heralded by Kant, Marx and others, understands spontaneity as a fundamental feature of agency that captures an agent’s ability to direct herself towards her own ends.

My dissertation is about how these two notions of spontaneity interact in cognition. I use attention as a vehicle for investigating how these notions collide with and complement each other. Attention is a suitable candidate because it contains active and passive components, has a robust empirical literature, and is the target of an increasing number of normative claims. What we notice, neglect, and do with our attention is the bedrock of mental action and the responsibility it entails.

I aim to address a series of questions.

1.    What type of control do we have over our attention?

2.    Which aspects of attention ground our responsibility practices that target attention?

3.    Do uncontrolled aspects of attention have normative significance?

4.    What is the unit of evaluation for attention norms?

5.    What impact, if any, does cognitive architecture have on how attention ought to be evaluated?

The first chapter (“Steering Salience”, R&R) focuses on salience in attention. Something is salient when it spontaneously grabs an agent’s attention. Salient attention is a clear case of where the above notions of spontaneity collide, as it has been the subject of two opposing intuitions across ethics, epistemology and cognitive science: 1) agents can be evaluated for what is salient to them and 2) salience is a passive process. Here I answer the second question above by way of the first. The aspects of attention that ground our responsibility practices are those that we have control over. The type of control is diachronic and cumulative.

A distinction between salient events—moments where attention is spontaneously captured—and salience patterns—an agent’s dispositions to attend—allows this diffusion strategy to work. Salient events are passive while salience patterns are evaluable. This is because salience patterns are subject to a learning process undergirded by how we respond to salient events. Part of what we are evaluating when we are judging an agent for finding something salient is past mental activity. This prior mental activity helps explain why agents find some feature to be salient. The process of learning salience patterns is informed by a series of past decisions that have systematic effects on an agent’s salience dispositions. This opens the door for a novel, non-deliberative form of activity to ground our evaluability practices, a position unavailable to both classic voluntarism and existing non-voluntarist alternatives to the responsibility for attitudes.

“Steering Salience” leaves questions alongside the answers it provides. Most prominently, how does the (ir)relevance judgement that informs an attentional response make them agential? And is there anything normatively significant about the passive, uncontrolled aspects of attention? These questions correspond with the second and third question above. The second chapter (“Creative Authorship”) uses the application case of creative action to build upon the account introduced in “Steering Salience” and address these further questions. 

Creative authorship is a philosophically puzzling instance of attribution that similarly pits the two notions of spontaneity against each other. Testimony from artists across time, medium and culture suggest that they are in some sense passive to the creative process. Yet our responsibility practices demand some attribution story that allows us to pin the outputs of a creative process to artists we praise and compensate. This is a particularly difficult case for my voluntarism given the embrace of passivity in testimony. While the account developed in the first chapter allows for pockets of passivity in the salient event, it does not imbue normative significance on those moments.

Unlike traditional approaches to solving the attribution problem of creativity that explain away the testimony of passivity, I embrace it. I argue that this persistent testimony reflects an aspect of agency that ought to figure more prominently into agency par excellence: receptivity, characterized broadly as an openness to the world. My account can uniquely take on this testimony as an insight into this specific agential ideal because it allows for pockets of passivity within the diachronic control story that grounds attribution claims. Answering the third question above: the uncontrolled aspects of attention have a normative significance insofar as they open the agent up to new and unforeseen content to figure into their guiding (aesthetic) normative standards.

This chapter doubles to expand of an underdeveloped aspect of my diachronic account of salience formation. On the learning model, the type of attentional action performed by an agent in response to a salient event is informed by relevance. These doxastic states (which I call aesthetic judgments) shape the internal normative standards that support attentional actions as actions. This development is significant as it provides tools for determining whether and when an artificial agent can be an author of a piece of art.

The third chapter (“Mavericks of the Mind”) provides a critical addition to my answer to the third question above: we have exploratory capacities that contribute to our practical development through spontaneously generating action tendencies that violate our guiding norms. I argue that intrusive thoughts have an agential function, deepening the normative significance of uncontrolled aspects of attention. Intrusive thoughts function to present agents with opportunities to reinforce and revise one’s guiding norms. We need checks like these because they can help us break away from how we previously framed a context, to remain attuned to shifting contexts and avoid local maxima. These exploratory capacities complement attitudes of receptivity to form a more complete account of the normative significance of uncontrolled aspects of the mind.

This chapter also furnishes an argument about the normative obligations an agent has towards these uncontrolled aspects of attention. Here, I depart from the previous two chapters which were concerned with establishing the normative grounds of those elements. Obligations are implied by these grounds, yet it is opaque what the agent can and ought to do with respect to spontaneous thoughts containing norm violations. Intrusive thoughts operate at the periphery of rationality, neither adherent to one’s guiding norms nor intentionally initiated.

I argue that the type of obligations one has towards these thoughts is at the capacity level, rather than the level of any individual thought. We ought to nurture our exploratory capacities to maintain (and improve) their functionality. Akin to an imperfect duty, a nurturing norm does not require the agent to respond any way to an individual intrusion; there is latitude in how one nurtures their exploratory capacity. Intrusive thoughts are therefore a tool for excavating an underappreciated imperfect duty to oneself: to nurture one’s exploratory capacities.

The fourth and final chapter (“Towards an Attentional Ecology”) aims to answer the final two questions listed above. The prior chapters implicitly argue that any attentional instance cannot be understood in isolation: it must be situated within the (agential) history that shaped it. This is formalized as a general problem for any theory that takes attention to be an evaluative target. The continuous nature of attention ushers in a permissibility problem for attention norms: it is never possible, in isolation, to determine whether an attentional instant is permissible with respect to an attention norm. In response, I propose that the unit of evaluation for attention norms ought to be dynamically holistic.

The contribution of this chapter is primarily methodological. I support this holistic claim about the unit of evaluation in attention norms by bolstering another implicit element of the previous chapters: the cognitive architecture of attention ought to influence our responsibility practices. I advocate for the indispensability of cognitive science in both demarcating the dynamic system we are evaluating with attention norms and informing the metrics we use for evaluating that system. Solving the permissibility problem for attention norms is dependent on this empirical work, dependent on understanding the descriptive nature of attention.