Sonia de Jager

This article explores the concept of active ignorance to elucidate the overlooked role of ignorance within the frameworks of predictive processing (PP) and active inference (AIF). We call this ignorance active because it refers to an operative refusal of learning that is actively maintained rather than a default or passive ignorance accompanying a lack of experience or education. Literature within PP and the AIF is beginning to criticize the view of the brain as an epistemic engine seeking to optimize its model of the world, and to look at the negative implications of the description of cognition within these frameworks. This paper seeks to contribute to this literature. Active ignorance is characterized as an ‘agno-inferential’ tendency (i.e., inference by ignorance) whereby a cognitive system reduces uncertainty by ignoring, denying or altering evidence that contrasts with deeply held beliefs, expectations, or hypotheses. We build upon analyses of (socio-cultural) ignorance in agnotology, epistemic injustice and critical race theory to articulate the concept of active ignorance. We then make explicit some of the implications of the AIF at the cultural niche-construction level by discussing how ignorance is a pervasive aspect of cognition structured according to its historical and political geography. We argue that the tacit conception of mind as progressing towards knowledge, implicit in many PP and AIF accounts, is inadequate. This conception of the mind can obstruct the consideration of persistently maintained false inferences, maladaptive features of socially-scaffolded cognition, and the identification of features that are favorable to specific types of agents.

This article presents an account of social power dynamics that centers on the notions of attention, control and scripts, elucidated through active inference as the evolution of shared policies with attractive tendencies. The acquisition and maintenance of social power depends on (changing) established habits: patterns such as languages, traditions, currencies, and other cultural norms—all of which will be framed in this article as scripts, because they represent communicable and iterable policies, or sequences of discrete steps. The first line of argument explored is social power as an effect of agents or groups exploring and exploiting how scripts control perspectives; i.e., attention. It will be argued that social power manifests through scripts in the capacity to modulate information-processing, and material conditions, through scripts as policies that organize attention and behavior. These processes can be understood as flows of information which can be modulated, and are therefore amenable to an information-theoretic, computational analysis. Observing the effects of scripts directing behavior as the basis of power and its material effects, leads to an argument about power as—effective and/or ever more possibilistic—computation, where computation is understood as a function occurring on a finite substrate that transforms an input into an output. Social empowerment is thus what grants agents more computing opportunities (as in: the concentration of attention and behavior as/or the transformation of matter), shaping the unfolding of future trajectories. Starting from the context of active inference, social power is therefore conceptualized as a function of variables including attention-attraction capabilities and information-processing efficiency, which together enable both script-creation and/or maintenance. Through an analysis of script-function, it is examined how power relations materialize through attention as information-processing, revealing straightforward connections between social habits, cognitive-behavioral patterns, and ensuing materiotechnical realities.

In active inference, mystical experiences (MEs) could be characterized as adaptive learning events, where the ordinary expectations of an agent become challenged or change permanently. In the context of psychedelic-induced MEs, these (temporary) changes in information processing have been attributed to increased brain entropy, the effects of which have been seen to reorganize deep perceptual-cognitive preferences (i.e. priors, e.g. an agent’s very sense of self). Adaptive (hyper)prior restructuring is an agent’s best survival strategy against environmental entropy: organisms survive by effectively predicting their future states. (Psychedelic) MEs, if characterized by increased brain entropy, may represent a state where sensory input and predictive output significantly overlap, leading to what is often phenomenally reported as a sense of “unity.” This possible alignment between sensorial entropy and entropic attunement could be understood evolutionarily as an adaptive strategy for the management of uncertainty, potentially explaining the profound nature of these experiences, where fundamental aspects of active inference come to a “halt,” delivering perplexity. This article explores how these experiences, framed as evolved allostatic adaptation, can be analyzed as information exchanges between systems where the discrepancies between the sensed and the predicted can be understood as radically altered, which we speculate is what phenomenally translates to cognitive effects such as higher ambiguity tolerance: a trademark of (psychedelic) MEs, but also death acceptance, aspects of dreams or psychosis, and comparable so-called “primary state” phenomena.

Semantic noise, the effect ensuing from the denotative and thus functional variability exhibited by different terms in different contexts, is a common concern in natural language processing (NLP). While unarguably problematic in specific applications (e.g., certain translation tasks), the main argument of this paper is that failing to observe this linguistic matter of fact as a generative effect rather than as an obstacle, leads to actual obstacles in instances where language model outputs are presented as neutral. Given that a common and long-standing challenge in NLP is the interpretation of ambiguous – i.e., semantically noisy – cases, this article focuses on an exemplar ambiguity-resolution task in NLP: the problem of anaphora in Winograd schemas. The main question considered is: to what extent is the standard approach to disambiguation in NLP subject to a stagnant “image of language”? And, can a transdisciplinary, dynamic approach combining linguistics and philosophy elucidate new perspectives on these possible conceptual shortcomings? In order to answer these questions we explore the term and concept of noise, particularly in its presentation as semantic noise. Owing to its definitional plurality, and sometimes even desirable unspecificity, the term noise is thus used as proof of concept for semantic generativity being an inherent characteristic in linguistic representation, and its concept is used to interrogate assumptions admitted in the resolution of Winograd schemas. The argument is speculative and theoretical in method, and the result is an analysis which provides an account of the fundamentally dialogical and necessarily open-ended effects of semantic noise in natural language.