Source-linked AI summary
Cognition and Reality
F. Tito Arecchi
TL;DR
The paper distinguishes apprehension from judgment and asks how linguistic comparison produces meanings beyond direct perceptual inference. It characterizes judgment as an inverse Bayes procedure and links the resulting circle and coil hermeneutics to creativity and scientific discovery. The paper concludes that repeated linguistic comparison can extract elements of reality, while creative language can also generate ambiguous situations without an external referent.
Problem
The paper addresses the overlooked distinction between emotion-linked apprehension and self-conscious judgment based on comparing successive linguistic apprehensions.
Method
The paper models apprehension as direct Bayes inference and judgment as inverse Bayes comparison of successive language-coded pieces retrieved from memory.
Results
The paper concludes that inverse Bayes language operations generate creative jumps, while repeated comparisons in scientific observation extract elements of reality.
Takeaways & Limitations
The circle operates within a pre-assigned ontology, whereas the coil progressively grasps new aspects through repeated interpretation.
Takeaways & Limitations
Emotion- or fMRI-based apprehension evidence is inadequate for explaining the judgment process, because judgment depends on contextual comparison between linguistic pieces.
Abstract
from arXiv · showhide
We discuss the two moments of human cognition, namely, apprehension (A), whereby a coherent perception emerges from the recruitment of neuronal groups, and judgment(B),that entails the comparison of two apprehensions acquired at different times, coded in a suitable language and retrieved by memory. (B) entails self-consciousness, in so far as the agent who expresses the judgment must be aware that the two apprehensions are submitted to his/her own scrutiny and that it is his/her task to extract a mutual relation. Since (B) lasts around 3 seconds, the semantic value of the pieces under comparison must be decided within that time. This implies a fast search of the memory contents. As a fact, exploring human subjects with sequences of simple words, we find evidence of a limited time window , corresponding to the memory retrieval of a linguistic item in order to match it with the next one in a text flow (be it literary, or musical,or figurative). While apprehension is globally explained as a Bayes inference, judgment tresults from an inverse Bayes inference. As a consequence, two hermeneutics emerge (called respectively circle and coil). The first one acts in a pre-assigned space of features. The second one provides the discovery of novel features, thus unveiling previously unknown aspects and hence representing the road to reality.
2-The brain operations - Role of homoclinic chaos
The paper describes neuronal activity through temporal and geometric chaos, spike synchronization, and competition among specialized cortical areas. These mechanisms connect individual firing to coordinated brain operations.
- 2-The brain operations - Role of homoclinic chaos: Geometric chaos describes divergence of dynamical trajectories caused by extreme sensitivity to initial conditions.The loss of initial information occurs over a time τ, whose inverse is the Kolmogorov entropy K.
- 2-The brain operations - Role of homoclinic chaos: Temporal chaos produces regular closed neuronal trajectories whose spike trains repeat at irregular time intervals.A single neuron emits spikes averaging 25 ms apart in the γ EEG band, with a 3 ms minimum separation.
- 2-The brain operations - Role of homoclinic chaos: Neurons communicate directly through axons or indirectly through local EEG potentials that alter the firing rate of distant neurons.Direct communication couples spike trains electrically, whereas indirect communication uses a signal χ generated by nearby neuronal activity.
- 2-The brain operations - Role of homoclinic chaos: Specialized cortical areas consist of nearby neurons coordinated around common tasks and communicating with other areas through EEG signals.Working regions can be visualized through their oxygenated-blood demand using f-MRI.
- 2-The brain operations - Role of homoclinic chaos: Two neuron groups receiving the same sensory stimulus compete under different top-down memory inputs, and the better-synchronized group produces the larger coherent signal.In Fig. 8, group I wins because its neurons sum coherently over Δt while group II remains less coordinated.
4-Perception as a Bayes inference
Perception is presented as a Bayes inference in which data update competing hypotheses through a model, selecting the most plausible interpretation and driving a reaction. Repeated updates progressively increase the plausibility of the selected hypothesis.
- 4-Perception as a Bayes inference: Bayesian perception begins with a manifold of hypotheses and uses a model to relate each hypothesis to possible data.The model is treated as an algorithm that generates different data for different hypotheses.
- 4-Perception as a Bayes inference: The observed data select h*, the hypothesis maximizing the posterior probability P(h|data).The selected hypothesis is the most plausible one under the measured data.
- 4-Perception as a Bayes inference: Bayes’ theorem computes the posterior as the prior probability of h multiplied by P(data|h) and divided by P(data).P(data|h) is the probability of the data conditioned on h and is called the model.
- 4-Perception as a Bayes inference: The perception procedure starts with an external stimulus and ends with a motor reaction.The selected hypothesis drives a suitable reaction within the depicted procedure.
- 4-Perception as a Bayes inference: Successive measurements and posterior updates reformulate new priors and climb toward increasing plausibility of h*.The paper compares this iterative procedure with Darwinian evolution through mutation and successive selection of the best-fit mutant.
4- Linguistic operations as inverse Bayes
The paper distinguishes apprehension from judgment, proposing that linguistic judgment compares successive text pieces through inverse Bayes rather than retrieving a fixed algorithm. This comparison requires self-consciousness, selects meanings by conformity with the next piece, and can generate new algorithms and semantic alternatives.
- Semantic complexity: Semantic complexity denotes the number of alternative meanings assigned to input data when cognition shifts among different algorithms.These alternatives correspond to moving among different probability-model “mountains,” unlike algorithmic complexity, which concerns the algorithm's bit length.
- Judgment and self-consciousness: Judgment requires self-consciousness because the agent must recognize the same examiner as comparing two non-simultaneous apprehensions and extracting their mutual relation.The comparison concerns apprehensions acquired at different times and coded in the same language.
- Inverse Bayes: Unlike apprehension, judgment does not presuppose an algorithm but builds a new one through comparison, which the paper associates with creativity and decisional freedom.The paper presents this model swap as a creative feature of language operations.
- Inverse Bayes: Judgment compares two successive pieces of a text retrieved through short-term memory, with inverse Bayes producing the appropriate conditional probability during comparison.The compared pieces may be literary, musical, or figurative, and the resulting operation generates a judgment.
- Limits of neural measures: Neural-correlate tests such as EEG and fMRI are presented as measures of neuronal recruitment and perceptual awareness, not self-conscious linguistic judgment.The paper therefore warns against treating emotional or apprehension-related measurements as evidence about judgment.
- Judgment and self-consciousness: Emotions support apprehension but are insufficient for judgment, whose meanings must be compared with the next word's code to select the best interpretation.This contrasts threshold-like competition in global workspace apprehension with comparison-based selection in judgment.
5-Two different hermeneutics, that is, interpretations of cognitive data
The paper contrasts circle hermeneutics, which repeatedly applies fixed meanings, with coil hermeneutics, which uses linguistic comparison to discover new aspects of an object. This expansion can also occur in interpersonal dialogue as both participants update their interpretations.
- Circle: Circle hermeneutics applies a fixed algorithm or ontology to generate knowledge within a pre-assigned space of features.Repeated reconsideration returns to the same memorized object, leaving no new insight.
- Coil: Coil hermeneutics revisits a text with provisional interpretations and progressively discovers new aspects and meanings.The process advances from A1 to A2 while identifying B2, B3, and later features of B.
- Coil: Language enables coil interpretation to make infinite use of finite resources without requiring a large amount of computational resources.The paper attributes this property to Humboldt’s account of language.
- Interpersonal dialogue: Coil hermeneutics also describes interpersonal dialogue, where changes in one person’s interpretation reflect the other person’s activity and relational readjustment.Both participants undergo corresponding hermeneutic updates during dialogical exchange.
6-Conclusions- Two aspects of linguistic creativity
Linguistic creativity can generate contradictory characters when language lacks an external referent, while comparisons of scientific observations can extract elements of reality. The paper proposes inverse-Bayes comparisons and paradigm shifts as a more efficient scientific program than fixed algorithms for increasingly complex systems.
- Linguistic creativity: Linguistic creation can produce ambiguous or mutually conflicting behaviors, exemplified by characters such as Ulysses and Don Quixote.The paper explains chimeras as arising from the lack of an external referent B.
- Scientific observation: Repeated linguistic comparison of scientific observations extracts elements of reality, unlike linguistic action that proceeds from a known piece toward an unknown one.The comparison concerns observed items, including successive pieces in scientific or poetic interpretation.
- Scientific program: As the observed world grows, fixed-algorithm science faces exponentially increasing computational size C and decreasing reliable prediction time τ, hence increasing Kolmogorov entropy K=1/τ.The paper frames this scaling problem as a consequence of applying a universal fixed algorithm to richer physical systems.
- Scientific program: Inverse-Bayes linguistic comparisons use non-algorithmic paradigm shifts to produce novel theories with low C and K, termed effective science.The paper gives Maxwell’s electromagnetic equations as an example of unifying previously separate phenomena.