Action Selection Theories

Lens adapted to this topic: Leading computational, behavioral, and neural frameworks

Mainstream research treats action selection as a functional problem: an agent must choose or organize behavior appropriate to its goals, internal condition, and environment. It does not require conscious deliberation. Accounts include competing affordances, inhibitory and priority mechanisms, basal-ganglia and cortical processes, behavioral preference factors, and computational models. Researchers disagree about implementation and the correct unit of selection, but commonly investigate how multiple possible behaviors are resolved into effective action.

• Action selection is the process of choosing what to do next from an agent’s available actions, given goals, internal state, and environmental context.
• Artificial-intelligence theories may use strictly sequential plans, while natural-intelligence models can treat behavior as distributed or dynamically emergent rather than as discrete chosen acts.
• The affordance-competition approach explains action selection in terms of multiple potential actions being represented and competing for control.
• Neuroscience reviews identify neural mechanisms that select among innate behaviors according to internal states and external conditions, while noting that these mechanisms remain poorly understood.
• Behavioral studies examine why some possible actions are chosen over other relevant or feasible actions, including in object-manipulation tasks.

Lens adapted to this topic: Challenges to selection-centered accounts of action and attention

Critical and outsider perspectives question whether every intentional action is fundamentally a solution to a selection problem. Critics argue that some intentional actions may be causally necessitated, that reflexes may also involve selection, and that making attention itself an action can create regress or agency problems. Challenges to attention-as-selection-for-action also argue that the theory risks mishandling attention capture and inattentive episodes, and may conflict with biased-competition accounts of visual attention.

• Some intentional actions may be causally necessitated, so critics argue they cannot all be explained as solutions to a choice among causally possible alternatives.
• The contrast between action and reflex may not establish that action must solve a selection problem or that reflexes do not solve one.
• If attention is itself an action, defining action through attention may generate a regress; if attention is a reflex, critics question what role remains for the agent.
• The claim that attention is necessarily selection for an actually occurring action may either exclude clear cases of attention capture or classify many apparently inattentive episodes as attentive.
• Critics argue that selection-for-action accounts face tensions with biased-competition theories of visual attention and with explaining how attending can itself count as an action.