Epistemic Feasibility and Interdependence in Normative Theorizing

If most agree that normative theorizing is constrained by an ‘ought implies can’ proviso and that feasibility consists in ‘overall human agential possibility’ (Valentini 2017: 24), widespread disagreement persists regarding different kinds of feasibility. One distinguishes between ‘hard’ and ‘soft’ limits on feasibility (Gilabert and Lawford-Smith 2012) but also between ‘binary’ and ‘scalar’ conceptions (Lawford-Smith 2013). Although philosophers have used such tools to explore logical, metaphysical, cultural and psychological species of feasibility, they have neglected epistemic feasibility. I attempt to close that gap by asking whether epistemic feasibility i.) sets stringent conditions on normative theorizing and ii.) takes priority over other feasibility conditions. In answer, I first elaborate an objection to normative theories presupposing expansive epistemic limits. I then show why that objection fails to heed the interdependent character of epistemic feasibility. The objection holds that there exists a disanalogy between epistemic feasibility and other species. Unlike institutional, cultural or motivational resources, human epistemic resources are fallible and fixed, leaving it unclear how and how far limits on cognitive and rational capacities are removable (Dipaolo 2019). The objection concludes that facts about cognitive and rational capacities must be taken for granted in normative theorizing. Evolutionary processes and AI advances are unpredictable in terms of correcting fully for cognitive and rational fallibility. Genetic engineering or neural reprogramming seem ethically fraught, unreasonable means for correction. Either way, human access to improved informational outputs would still be mediated by fallible epistemic processes. Wherefore two refined provisos for normative theorizing. First, if ‘ought implies epistemic can’, theories presupposing expansive epistemic limits cannot guide individual or collection action. This discounts such approaches as theories of justice supposing impossible perspectives or perfect information, theories of governance entailing centrally planned economies or global states with perfect information, etc. Second, if ‘epistemic ought implies can’, philosophers must avoid elaborating and justifying normative theories presupposing expansive epistemic limits. Nevertheless, the objection fails by focusing exclusively on individual epistemic feasibility and ignoring epistemic interdependence qua distributed epistemic feasibility. Philosophers understand epistemic limits expansively because the world (as accessible to humans) resists representation in a single picture of the world. Those limits necessitate abstraction from certain real-world conditions to find a model suitable for a specific problem. Comparing different models – each suited for a specific problem – may generate a more adequate overall picture. If epistemic limits condition working with multiple pictures (i.e. epistemically perfect agents could not) and multiple pictures are necessary to navigate the environment, epistemic feasibility is distributed across interdependent epistemic agents. Philosophers count on others to assume away different epistemic limits which complement her normative theorizing’s assumptions. Analyzed as a set, their assumptions cancel one another out, and epistemic feasibility obtains collectively. Thus, epistemic limits are assumed away because epistemic feasibility itself requires expansive limits. Although there may be a disanalogy between epistemic feasibility and other species, the objection is mistaken about its nature. Epistemic feasibility just is a matter of epistemic interdependence.