arXiv subjects

Giulio Ruffini

Publications and source records attributed to Giulio Ruffini.

At least 19 recordsLinked to original sources

arXiv2026

Large-Language Models as a Cognitive Virus

Large-language models (LLMs) are rapidly becoming part of human culture, reshaping how information is produced, transmitted, and used. Here we propose that their diffusion can be understood through a viral analogy, with LLM use spreading through populations, becoming embedded in cognitive and cultural practices. We model transitions among uncoupled, coupled, and persistently dependent users, and show that the interplay between social transmission, recovery, and collective reinforcement can generate tipping points and technological lock-in. A central consequence is the possibility of runaway dynamics: once a critical threshold is crossed, small increases in adoption can trigger rapid population-level shifts toward persistent dependence, with abrupt losses in cognitive competence. The same framework, however, identifies conditions for cognitive immunization, based on reducing transmission and facilitating reversibility. Our results highlight how LLM adoption may involve nonlinear collective transitions with important consequences for cognitive autonomy.

physics.soc-ph↗

arXiv2025

Rosetta Stone of Neural Mass Models

Brain dynamics dominate every level of neural organization -- from single-neuron spiking to the macroscopic waves captured by functional magnetic resonance imaging (fMRI), magnetoencephalography (MEG), and electroencephalography (EEG) -- yet the mathematical tools used to interrogate those dynamics remain scattered across a patchwork of traditions. Neural mass models (NMMs) (aggregate neural models) provide one of the most popular gateways into this landscape, but their sheer variety -- spanning lumped parameter models, firing-rate equations, and multi-layer generators -- demands a unifying framework that situates diverse architectures along a continuum of abstraction and biological detail. Here, we start from the idea that oscillations originate from a simple push-pull interaction between two or more neural populations. We build from the undamped harmonic oscillator and, guided by a simple push-pull motif between excitatory and inhibitory populations, climb a systematic ladder of detail. Each rung is presented first in isolation, next under forcing, and then within a coupled network, reflecting the progression from single-node to whole-brain modeling. By transforming a repertoire of disparate formalisms into a navigable ladder, we hope to turn NMM choice from a subjective act into a principled design decision, helping both theorists and experimentalists translate between scales, modalities, and interventions. In doing so, we offer a Rosetta Stone for brain oscillation models -- one that lets the field speak a common dynamical language while preserving the dialectical richness that fuels discovery.

q-bio.NC↗

arXiv2025

Compositional Symmetry as Compression: Lie Pseudogroup Structure in Algorithmic Agents

In the algorithmic (Kolmogorov) view, agents are programs that track and compress sensory streams using generative programs. We propose a framework where the relevant structural prior is simplicity (Solomonoff) understood as \emph{compositional symmetry}: natural streams are well described by (local) actions of finite-parameter Lie pseudogroups on geometrically and topologically complex low-dimensional configuration manifolds (latent spaces). Modeling the agent as a generic neural dynamical system coupled to such streams, we show that accurate world-tracking imposes (i) \emph{structural constraints} -- equivariance of the agent's constitutive equations and readouts -- and (ii) \emph{dynamical constraints}: under static inputs, symmetry induces conserved quantities (Noether-style labels) in the agent dynamics and confines trajectories to reduced invariant manifolds; under slow drift, these manifolds move but remain low-dimensional. This yields a hierarchy of reduced manifolds aligned with the compositional factorization of the pseudogroup, providing a geometric account of the ``blessing of compositionality'' in deep models. We connect these ideas to the Spencer formalism for Lie pseudogroups and formulate a symmetry-based, self-contained version of predictive coding in which higher layers receive only \emph{coarse-grained residual transformations} (prediction-error coordinates) along symmetry directions unresolved at lower layers.

cs.LG↗

arXiv2025

The Algorithmic Regulator

The regulator theorem states that, under certain conditions, any optimal controller must embody a model of the system it regulates, grounding the idea that controllers embed, explicitly or implicitly, internal models of the controlled. This principle underpins neuroscience and predictive brain theories like the Free-Energy Principle or Kolmogorov/Algorithmic Agent theory. However, the theorem is only proven in limited settings. Here, we treat the deterministic, closed, coupled world-regulator system $(W,R)$ as a single self-delimiting program $p$ via a constant-size wrapper that produces the world output string~$x$ fed to the regulator. We analyze regulation from the viewpoint of the algorithmic complexity of the output, $K(x)$. We define $R$ to be a \emph{good algorithmic regulator} if it \emph{reduces} the algorithmic complexity of the readout relative to a null (unregulated) baseline $\varnothing$, i.e., \[ \Delta = K\big(O_{W,\varnothing}\big) - K\big(O_{W,R}\big) > 0. \] We then prove that the larger $\Delta$ is, the more world-regulator pairs with high mutual algorithmic information are favored. More precisely, a complexity gap $\Delta > 0$ yields \[ \Pr\big((W,R)\mid x\big) \le C\,2^{\,M(W{:}R)}\,2^{-\Delta}, \] making low $M(W{:}R)$ exponentially unlikely as $\Delta$ grows. This is an AIT version of the idea that ``the regulator contains a model of the world.'' The framework is distribution-free, applies to individual sequences, and complements the Internal Model Principle. Beyond this necessity claim, the same coding-theorem calculus singles out a \emph{canonical scalar objective} and implicates a \emph{planner}. On the realized episode, a regulator behaves \emph{as if} it minimized the conditional description length of the readout.

cs.CC↗

arXiv2022

Comparison between an exact and a heuristic neural mass model with second order synapses

Neural mass models (NMMs) are designed to reproduce the collective dynamics of neuronal populations. A common framework for NMMs assumes heuristically that the output firing rate of a neural population can be described by a static nonlinear transfer function (NMM1). However, a recent exact mean-field theory for quadratic integrate-and-fire (QIF) neurons challenges this view by showing that the mean firing rate is not a static function of the neuronal state but follows two coupled nonlinear differential equations (NMM2). Here we analyze and compare these two descriptions in the presence of second-order synaptic dynamics. First, we derive the mathematical equivalence between the two models in the infinitely slow synapse limit, i.e., we show that NMM1 is an approximation of NMM2 in this regime. Next, we evaluate the applicability of this limit in the context of realistic physiological parameter values by analyzing the dynamics of models with inhibitory or excitatory synapses. We show that NMM1 fails to reproduce important dynamical features of the exact model, such as the self-sustained oscillations of an inhibitory interneuron QIF network. Furthermore, in the exact model but not in the limit one, stimulation of a pyramidal cell population induces resonant oscillatory activity whose peak frequency and amplitude increase with the self-coupling gain and the external excitatory input. This may play a role in the enhanced response of densely connected networks to weak uniform inputs, such as the electric fields produced by non-invasive brain stimulation.

q-bio.NC↗

arXiv2017

Lempel-Zip Complexity Reference

The aim of this note is to provide some reference facts for LZW---mostly from Thomas and Cover \cite{Cover:2006aa} and provide a reference for some metrics that can be derived from it. LZW is an algorithm to compute a Kolmogorov Complexity estimate derived from a limited programming language that only allows copy and insertion in strings (not Turing complete set). Despite its delightful simplicity, it is rather powerful and fast. We then focus on definitions of LZW derived complexity metrics consistent with the notion of descriptive length, and discuss different normalizations, which result in a set of metrics we call $\rho_0$, $\rho_1$ and $\rho_2$, in addition to the Description Length $l_{LZW}$ and the Entropy Rate.

cs.IT↗

arXiv2016

Models, networks and algorithmic complexity

I aim to show that models, classification or generating functions, invariances and datasets are algorithmically equivalent concepts once properly defined, and provide some concrete examples of them. I then show that a) neural networks (NNs) of different kinds can be seen to implement models, b) that perturbations of inputs and nodes in NNs trained to optimally implement simple models propagate strongly, c) that there is a framework in which recurrent, deep and shallow networks can be seen to fall into a descriptive power hierarchy in agreement with notions from the theory of recursive functions. The motivation for these definitions and following analysis lies in the context of cognitive neuroscience, and in particular in Ruffini (2016), where the concept of model is used extensively, as is the concept of algorithmic complexity.

cs.LG↗

arXiv2015

Application of the reciprocity theorem to EEG inversion and optimization of EEG-driven transcranial current stimulation (tCS, including tDCS, tACS, tRNS)

Multichannel transcranial current stimulation (tCS) systems offer the possibility of EEG-guided optimized, non-invasive brain stimulation. In this brief technical note I explain how it is possible to use tCS electric field realistic brain model to create a forward "lead-field" matrix and, from that, an EEG inverter for cortical mapping. Starting from EEG I show how to generate 2D cortical surface dipole fields that could produce the observed EEG electrode voltages. The main tool is the reciprocity theorem derived by Helmholtz. The application of reciprocity for the generation of a forward mapping matrix (lead field matrix as is sometimes known) is well known [Rush and Driscoll, 1969], but here we will use it in combination with the realistic head models of [Miranda et al 2013] to provide cortical mapping solutions compatible with realistic head model tCS optimization. I also provide a generalization of the reciprocity theorem [Helmholtz 1853] to the case of multiple electrode contact points and dipole sources, and discuss its uses in non-invasive brain stimulation based on EEG. This, as far as I know, is a novel result. Applications are discussed.

physics.bio-ph↗

arXiv2009

Reality as Simplicity

The aim of this paper is to study the relevance of simplicity and its formal representation as Kolmogorov or algorithmic complexity in the cognitive sciences. The discussion is based on two premises: 1) all human experience is generated in the brain, 2) the brain has only access to information. Taken together, these two premises lead us to conclude that all the elements of what we call `reality' are derived mental constructs based on information and compression, i.e., algorithmic models derived from the search for simplicity in data. Naturally, these premises apply to humans in real or virtual environments as well as robots or other cognitive systems. Based on this, it is further hypothesized that there is a hierarchy of processing levels where simplicity and compression play a major role. As applications, I illustrate first the relevance of compression and simplicity in fundamental neuroscience with an analysis of the Mismatch Negativity paradigm. Then I discuss the applicability to Presence research, which studies how to produce real-feeling experiences in mediated interaction, and use Bayesian modeling to define in a formal way different aspects of the illusion of Presence. The idea is put forth that given alternative models (interpretations) for a given mediated interaction, a brain will select the simplest one it can construct weighted by prior models. In the final section the universality of these ideas and applications in robotics, machine learning, biology and education is discussed. I emphasize that there is a common conceptual thread based on the idea of simplicity, which suggests a common study approach.

physics.gen-ph↗

arXiv2008

Soil Moisture Monitorization Using GNSS Reflected Signals

The use of GNSS signals as a source of opportunity for remote sensing applications, GNSS-R, has been a research area of interest for more than a decade. One of the possible applications of this technique is soil moisture monitoring. The retrieval of soil moisture with GNSS-R systems is based on the variability of the ground dielectric properties associated to soil moisture. Higher concentrations of water in the soil yield a higher dielectric constant and reflectivity, which incurs in signals that reflect from the Earth surface with higher peak power. Previous investigations have demonstrated the capability of GPS bistatic scatterometers to obtain high enough signal to noise ratios in order to sense small changes in surface reflectivity. Furthermore, these systems present some advantages with respect to others currently used to retrieve soil moisture. Upcoming satellite navigation systems, such as the European Galileo, will represent an excellent source of opportunity for soil moisture remote sensing for various reasons. First, the existence of pilot signals will provide the possibility to extend coherent integration times, which will contribute to the increase of received signals SNR. In addition, the availability of Galileo L1 and L5 signals will allow the multi-spectral analysis of the reflected signals and the development of inversion models which will be able to account more precisely for adverse effects, such as surface roughness and vegetation canopy. In this paper we present some of the recent theoretical work and experiments carried out at Starlab focusing on the development of dedicated Soil Moisture GNSS-R systems.

physics.geo-ph↗

arXiv2007

Information, complexity, brains and reality (Kolmogorov Manifesto)

I discuss several aspects of information theory and its relationship to physics and neuroscience. The unifying thread of this somewhat chaotic essay is the concept of Kolmogorov or algorithmic complexity (Kolmogorov Complexity, for short). I argue that it is natural to interpret cognition as the art of finding algorithms that apprach the Solomonoff-Kolmogorov-Chaitin (algorithmic) Complexity limit with appropriate tradeoffs. In addition, I claim that what we call the universe is an interpreted abstraction--a mental construct--based on the observed coherence between multiple sensory input streams and our own interactions. Hence, the notion of Universe is itself a model.

physics.gen-ph↗

arXiv2006

A dry electrophysiology electrode using CNT arrays

We describe the concept of a dry electrode sensor for biopotential measurement applications (ENOBIO) designed to eliminate the noise and inconvenience associated to the use of electrolytic gel. ENOBIO uses nanotechnology to remove gel-related noise, as well as maintaining a good contact impedance to minimise interference noise. The contact surface of the electrode will be covered with an array/forest of carbon nanotubes and will also be tested with an Ag/AgCl coating to provide ionic-electronic transduction. The nanotubes are to penetrate the outer layers of the skin, the Stratum Corneum, improving electrical contact. We discuss requirements, skin properties, nanotube penetration and transduction, noise sources, prototype design logic and biocompatibility. A future paper will report test results.

physics.med-ph↗

arXiv2006

ENOBIO - First tests of a dry electrophysiology electrode using carbon nanotubes

We describe the development and first tests of Enobio, a dry electrode sensor concept for biopotential applications. In the proposed electrodes, the tip of the electrode is covered with a forest of multi-walled CNTs that can be coated with Ag/AgCl to provide ionic-electronic transduction. The CNT brush-like structure is to penetrate the outer layers of the skin improving electrical contact as well as increae the contact surface area. In this paper, we report the results of the first tests of this concept -- immersion on saline solution and pig skin signal detection. These indicate performance on a par with state of the art research-oriented wet electrodes.

physics.med-ph↗

arXiv2002

Ionospheric (H-atom) Tomography: a Feasibility Study using GNSS Reflections

In this report we analyze the feasibility of ionospheric monitoring using GNSS technology. The focus will be on the use of LEO GNSS data, exploiting GNSS Reflections, Navigation and Occultation TEC measurements. In order to attack this question, we have simulated GNSS ionospheric TEC data as it would be measured from a polar LEO (exploiting Navigation, Occultation and Reflection TEC data) and IGS ground stations, through the use of a climatic ionospheric model (we have explored both NeQuick and PIM). We have then developed a new tomographic approach inspired on the physics of the hydrogen atom, which has been compared to previous successful but somewhat awkward methods (using a voxel representation) and employed to retrieve the Electronic Density field from the simulated TEC data. These tomographic inversion results using simulated data demonstrate the significant impact of GNSS-R and GNSS-NO data: 3D ionospheric Electron Density fields are retrieved over the oceans quite accurately, even as, in the spirit of this initial study, the simulation and inversion approaches avoided intensive computation and sophisticated algorithmic elements (spatio-temporal smoothing). We conclude that GNSS-R data can contribute significantly to the GIOS (Global/GNSS Ionospheric Observation System).

physics.ao-ph↗

arXiv2002

PARIS Altimetry with L1 Frequency Data from the Bridge 2 Experiment

A portion of 20 minutes of the GPS signals collected during the Bridge 2 experimental campaign, performed by ESA, have been processed. An innovative algorithm called Parfait, developed by Starlab and implemented within Starlab's GNSS-R Software package STARLIGHT (STARLab Interferometric Gnss Toolkit), has been successfully used with this set of data. A comparison with tide values independently collected and with differential GPS processed data has been performed. We report a successful PARIS phase altimetric measure of the Zeeland Brug over the sea surface with a rapidly changing tide, with a precision better than 2 cm.

physics.ao-ph↗

arXiv2002

Spherical Harmonics Interpolation, Computation of Laplacians and Gauge Theory

The aim in this note is to define an algorithm to carry out minimal curvature spherical harmonics interpolation, which is then used to calculate the Laplacian for multi-electrode EEG data analysis. The approach taken is to respect the data. That is, we implement a minimal curvature condition for the interpolating surface subject to the constraints determined from the multi-electrode data. We implement this approach using spherical harmonics interpolation. In this elegant example we show that minimization requirement and constraints complement each other to fix all degrees of freedom automatically, as occurs in gauge theories. That is, the constraints are respected, while only the orthogonal subspace minimization constraints are enforced. As an example, we discuss the application to interpolate control data and calculate the temporal sequence of laplacians from an EEG Mismatch Negativity (MMN) experiment (using an implementation of the algorithm in IDL).

physics.med-ph↗

arXiv2001

Stationary Phase in Coherent State Path Integrals

In applying the stationary phase approximation to coherent state path integrals a difficulty occurs; there are no classical paths that satisfy the boundary conditions of the path integral. Others have gotten around this problem by reevaluating the action. In this work it is shown that it is not necessary to reevaluate the action because the stationary phase approximation is applicable so long as the path, about which the expansion is performed, satisfies the associated Lagrange's equations of motion. It is not necessary for this path to satisfy the boundary conditions in order to apply the stationary phase approximation.

math-ph↗

arXiv1999

Analysis of Water Vapor spatio-temporal structure over the Madrid Area using GPS data

We have analyzed Zenith Wet Delay (ZWD) time series from an experiment over the Madrid (Spain) area obtained from 5 GPS receivers using two different techniques. In the first case a delay correlation analysis of the ZWD time-series has been carried out. We show that for this small network (with a spatial scale of less than 100 km) the correlation between the time series is very strong, and that using windowing techniques a reliable correlation delay time series can be produced for each pair of sites (10 such pairs are available). We use this delay time series together with a frozen flow model to estimate the velocity of a passing front, and compare the results to meteorological data and Numerical Weather Prediction output, showing good agreement. In the second approach, the data is analyzed using Empirical Orthogonal Functions. We demonstrate that the temporally demeaned and normalized analysis yields information about the passing of fronts, while the spatially demeaned data yields orographic information. A common second mode highlights the underlying wave behavior.

physics.ao-ph↗