Information-based Explanations in Cognitive Science

ABSTRACT:

Cognitive science currently operates as a “Babel tower” of disjointed disciplines and vocabularies, lacking a unified conceptual foundation despite a surface-level consensus on “information processing.” While the Information Processing Paradigm (IPP) remains a central pillar, it is increasingly under siege from skeptics such as radical enactivism and dynamic systems theorists, who challenge the necessity of internal representations. This dissertation defends a refined “neo-IPP” as the most robust explanatory strategy in the sciences of the mind, arguing that the information-based paradigm can be rescued through a rigorous, mechanistically grounded account of neural vehicles. Adopting the stance of “philosophy in/of cognitive science,” the work functions not only as a theoretical meta-analysis but as an active component of the explanatory project itself, contributing conceptual tools and formalisms directly applicable to neuroscientific practice.

The thesis consists of two main parts:

Theoretical Framework (Part 1)

The inquiry begins in Chapter 1 with a comprehensive audit of existing information theories, establishing a critical distinction between “a-information” (observer-dependent statistical measures) and “f-information” (system-internal functional substrates). This resolves the “hydraulic fallacy”—the error of treating information as a fluid-like substance rather than a structural property. Chapter 2 analyzes representations through the “I+X strategy,” examining how information becomes representation when supplemented with additional constitutive properties. The chapter identifies a substantive set of requirements for representational processes, including non-ubiquity, decouplability, portability, structural mirroring, and the possibility of misrepresentation. Chapter 3 examines representational format—a category frequently neglected in philosophical literature—filling a critical gap by analyzing how metabolic and computational constraints shape the structure of neural codes. Chapter 4 focuses on content, introducing the concept of the “consumer” to address critiques regarding the “code metaphor,” demonstrating that meaning requires a system that utilizes the signal. This theoretical grounding receives a concrete biological implementation in the “two-point processor” model, which identifies the biophysical architecture of the pyramidal neuron as the “mechanical judge” of informational content. Chapter 5 synthesizes these foundational concepts into the integrated “neo-IPP framework” and introduces the “explanatory cascade”—an explanatory strategy that marries information, representation, format and content in a naturalistic and causally-grounded way.

Empirical Stress Tests (Part 2)

The second half of the dissertation subjects this framework to empirical “stress tests” designed as publishable contributions to cognitive science. Chapter 6 analyzes how biological and computational constraints drive compression as representational efficacy that enables naturalistic abstraction and supports flexible and general-domain behaviour, framing compression as a semantic act of prioritizing verisimilar features. Chapter 7 utilizes Representational Similarity Analysis (RSA) within the “correspondence network framework” to argue that structural correspondences—or infomorphisms—allow the brain to extract meaning by matching geometries across disparate substrates. Finally, Chapter 8 defines the conditions for genuine generativity, distinguishing the genuine extrapolative capabilities of the neo-IPP from the “stochastic parroting” of simpler models. Ultimately, the dissertation provides a unified explanatory strategy—from causal grounding and structural compression to semantic validation—concluding that the informational picture remains the indispensable explanation for the flexible, “representation-hungry” nature of high-level cognition.

Full text available below.

Recommended citation: Mamak, W. (2026). Information-based Explanations in Cognitive Science, PhD Dissertation, Polish Academy of Sciences.
Download Paper