Revising Bloom's Taxonomy for Dual-Mode Cognition in Human-AI Systems: The Augmented Cognition Framework

TL;DR

This paper introduces the Augmented Cognition Framework (ACF), a revised taxonomy for human-AI systems that distinguishes between individual and distributed cognition modes and includes a governance level for managing mode-switching.

Contribution

It proposes a novel dual-mode taxonomy with explicit assessment targets and a governance level, addressing limitations of previous Bloom's revisions in AI-integrated cognition.

Findings

ACF generates assessable outcomes for both cognition modes and governance.

The framework explicitly models the dependency between individual and distributed cognition.

Systematic comparison shows ACF's unique utility over existing taxonomies.

Abstract

As artificial intelligence (AI) models become routinely integrated into knowledge work, cognitive acts increasingly occur in two distinct modes: individually, using biological resources alone, or distributed across a human-AI system. Existing revisions to Bloom's Taxonomy treat AI as an external capability to be mapped against human cognition rather than as a driver of this dual-mode structure, and thus fail to specify distinct learning outcomes and assessment targets for each mode. This paper proposes the Augmented Cognition Framework (ACF), a restructured taxonomy built on three principles. First, each traditional Bloom level operates in two modes (Individual and Distributed) with mode-specific cognitive verbs. Second, an asymmetric dependency relationship holds wherein effective Distributed cognition typically requires Individual cognitive foundations, though structured scaffolding can in some cases reverse this sequence. Third, a seventh level, Orchestration, specifies a governance capacity for managing mode-switching, trust calibration, and partnership optimization. We systematically compare existing AI-revised taxonomies against explicit assessment-utility criteria and show, across the frameworks reviewed, that ACF uniquely generates assessable learning outcomes for individual cognition, distributed cognition, and mode-governance as distinct targets. The framework addresses fluent incompetence, the central pedagogical risk of the AI era, by making the dependency relationship structurally explicit while accommodating legitimate scaffolding approaches.