The Background (Searle)

In the AI Story

Without the Background, representations cannot be interpreted. Interpretation requires the interpretive capacity that the Background provides. A sentence is a string of symbols; the Background is what makes the string mean something. Remove the Background, and the symbols revert to syntax — formal objects with no semantic content, manipulable by rules but incomprehensible to the manipulator. This is precisely the situation of the person in the Chinese Room: she manipulates symbols according to rules, but she lacks the Background that would allow her to interpret them.

The relevance to artificial intelligence is immediate. A large language model is trained on text produced by beings that possess Backgrounds. The training data encodes, in its statistical patterns, the traces that Background knowledge leaves in language — the associations, contexts, implicit assumptions that reflect embodied engagement with physical and social worlds. The model learns these patterns. It learns that "cut the cake" typically appears in contexts involving celebrations and knives, that "cut the grass" typically appears in contexts involving yards and mowers. But learning the statistical associations that Background knowledge produces is not the same as possessing Background knowledge. The map is not the territory. The representation of a capacity is not the capacity itself.

Understanding grounded in Background is robust — it generalizes to novel situations, handles ambiguity, catches errors that violate the implicit physics or social logic the Background encodes. Understanding without Background is brittle — it works within the patterns the training data covers and fails, sometimes catastrophically, outside them. The Deleuze failure documented in The Orange Pill is a failure of Background. The model learned statistical associations between "flow," "smooth," and "Deleuze." It did not possess the philosophical Background — the years of reading, arguing, misunderstanding, rereading, and gradually building embodied familiarity with a body of thought — that would have revealed the associations as superficial.

Searle's Background intersects with the arguments about depth and friction that Byung-Chul Han's philosophy raises. Han argues that removing friction produces smoothness — the aesthetic of surfaces without depth. Searle's Background provides the mechanism. Background knowledge is built through friction — through the embodied, time-consuming, often uncomfortable process of engaging with the world directly, making mistakes, recovering, and depositing layers of understanding that accumulate into competence. When Claude takes over the debugging or drafts the brief, the output may be correct, but the Background the human would have built through producing the output is not built. The output is extracted without the experiential deposit.

Origin

Searle developed the concept of the Background systematically in Intentionality (1983) and extended it in The Rediscovery of the Mind (1992) and later works. The concept has roots in Heidegger's analysis of being-in-the-world and Wittgenstein's discussions of forms of life, though Searle gave it a distinctive formulation tied to his theory of intentionality.

The Background was Searle's response to a specific problem: how do intentional states get their specific content? The literal meaning of a sentence does not determine its conditions of satisfaction; the same sentence can mean different things in different contexts. What supplies the contextual content? Searle's answer: the Background — the non-representational know-how that enables us to understand what counts as cutting the cake versus cutting the grass.

Key Ideas

Non-representational. The Background is not a set of beliefs, rules, or propositions. It is the capacity that makes belief, rule-following, and proposition-processing possible. It cannot be fully stated because stating would make it representational.

Embodied. Much of the Background is physical — bodily skills, proprioceptive know-how, the felt sense of how objects behave. These capacities live in the body; they are not accessible to a system without a body.

Deposited through friction. Background knowledge accumulates through sustained embodied engagement with resistant material. The senior engineer's debugging intuition, the lawyer's case-law sense, the philosopher's familiarity with an author — these are built through years of friction, and they cannot be transferred by documentation.

The interpretability gap. Robust understanding requires Background; without it, understanding is brittle. AI systems that lack Background fail unpredictably when situations deviate from training distributions in ways that violate implicit physics or social logic.

The developmental problem. When AI takes over the activities through which Background is built, the Background-building stops. The outputs continue, but the capacity to evaluate them degrades because the capacity was built through the activity the AI has absorbed.