arXiv subjects

Conor Houghton

Publications and source records attributed to Conor Houghton.

At least 19 recordsLinked to original sources

arXiv2026

Traceable Trust for action-ready artificial intelligence in bioscience

Artificial intelligence (AI) is becoming part of the working infrastructure of the biosciences. AI models can predict biomolecular structures, design proteins, rank variants, annotate images, recommend strains and optimise experimental conditions. We argue that the decision to use an AI output to guide laboratory action is a key juncture for trustworthy research and should follow a defined, reviewable process. We propose Traceable Trust as a proportionate assessment-and-design framework for this output-to-action boundary. It asks what evidence supports the output, what capability is being claimed, what agency has been delegated, what threshold authorises action, who can override it and how outcomes inform later decisions. We illustrate the framework through three case studies spanning ecosystem resources, project design and laboratory action. Together, the cases show how trust can be documented where AI outputs begin to shape scientific work.

cs.CY↗

arXiv2026

Approximate Muon with low-rank adapters

The Muon optimizer shows clear benefits versus alternatives when pretraining neural networks. However, it is used less frequently for parameter-efficient fine-tuning (PEFT). One potential reason is that the most common PEFT method, LoRA, does not naturally combine with Muon since it is not mathematically possible to orthogonalize the weight update given by a low-rank parameterization. In this paper, we address this issue by approximating the solution to a relaxed Muon objective in the low-rank setting via linearization and then least-squares. We provide an efficient implementation that uses matmul operations only, as opposed to more complex linear algebra decomposition routines. Our method, sMuon (small Muon), performs favourably across SFT and a ReLoRA pretraining experiment. While results are model- and eval-dependent, we find overall that using Muon for low-rank fine-tuning provides moderate performance improvements.

cs.LG↗

arXiv2026

Spatial community structure impedes language amalgamation in a population-based iterated learning model

The iterated learning model is an agent-based model of language evolution notable for demonstrating the emergence of compositional language. In its original form, it modelled language evolution along a single chain of teacher-pupil interactions; here we modify the model to allow more complex patterns of communication within a population and use the extended model to quantify the effect of within-community and between-community communication frequency on language development. We find that a small amount of between-community communication can lead to population-wide language convergence but that this global language amalgamation is more difficult to achieve when communities are spatially embedded.

cs.MA↗

arXiv2026

Evolving language compositionality in a frequency-structured meaning space

The iterated learning model was introduced to investigate language evolution: the way in which the characteristic properties of human languages have been shaped, at least partly, by repeated transmission from one language user to another. The key finding is that language compositionality can arise spontaneously as a consequence of language being passed repeatedly through a language learning bottleneck. Here we explore how changing the frequency of different meanings, so that some meanings occur much more frequently than others, affects the character of its compositionality. We find that, as observed in natural languages, high-frequency meanings can escape the pressure to conform to the grammar that characterizes lower-frequency meanings. However, when the frequency structure is instead imposed on parts rather than on whole meaning vectors, the language fails to transmit across generations. This occurs despite the fact that the most frequent elements are reliably learned. These results suggest that frequency can shape emergent linguistic structure only when the frequency distribution is defined over form-meaning units that learners can acquire holistically. When frequency is instead distributed over smaller units, it fails to support the relational structure required for compositional generalisation, thereby preventing stable language transmission.

cs.CL↗

arXiv2026

Energy budgets govern synaptic precision and its regulation during plasticity

Synaptic transmission must balance the need for reliable signalling against the metabolic cost of achieving that reliability. How energetic constraints shape synaptic precision and its regulation during plasticity remains unclear. Here we develop an energy--constrained framework in which synapses minimise postsynaptic response variance subject to a fixed mean and an effective energy budget. Combinations of candidate physiological costs are used to estimate an energy cost for synaptic transmission; this cost is then inferred from quantal statistics. Analysing five published pre- and post-plasticity datasets, we find that observed synaptic mean--variance pairs cluster near a minimal-energy boundary, indicating that precision is limited by energetic availability. Model comparison identifies a dominant calcium pump-like cost paired with a smaller vesicle turnover-like cost, yielding a separable precision--energy relationship, $σ^{-2} \propto E^5$. We further show that plasticity systematically updates synaptic energy budgets according to the scale-free magnitude of mean change, enabling accurate prediction of post-plasticity variance from energy allocation alone. These results provide direct experimental support for the hypothesis that synaptic precision is governed by energy budgets, establishing energy allocation as a fundamental principle linking metabolic constraints, synaptic reliability, and plasticity.

q-bio.NC↗

arXiv2026

Image, Word and Thought: A More Challenging Language Task for the Iterated Learning Model

The iterated learning model simulates the transmission of language from generation to generation in order to explore how the constraints imposed by language transmission facilitate the emergence of language structure. Despite each modelled language learner starting from a blank slate, the presence of a bottleneck limiting the number of utterances to which the learner is exposed can lead to the emergence of language that lacks ambiguity, is governed by grammatical rules, and is consistent over successive generations, that is, one that is expressive, compositional and stable. The recent introduction of a more computationally tractable and ecologically valid semi supervised iterated learning model, combining supervised and unsupervised learning within an autoencoder architecture, has enabled exploration of language transmission dynamics for much larger meaning-signal spaces. Here, for the first time, the model has been successfully applied to a language learning task involving the communication of much more complex meanings: seven-segment display images. Agents in this model are able to learn and transmit a language that is expressive: distinct codes are employed for all 128 glyphs; compositional: signal components consistently map to meaning components, and stable: the language does not change from generation to generation.

cs.CL↗

arXiv2025

Jacobian Sparse Autoencoders: Sparsify Computations, Not Just Activations

Sparse autoencoders (SAEs) have been successfully used to discover sparse and human-interpretable representations of the latent activations of LLMs. However, we would ultimately like to understand the computations performed by LLMs and not just their representations. The extent to which SAEs can help us understand computations is unclear because they are not designed to "sparsify" computations in any sense, only latent activations. To solve this, we propose Jacobian SAEs (JSAEs), which yield not only sparsity in the input and output activations of a given model component but also sparsity in the computation (formally, the Jacobian) connecting them. With a naïve implementation, the Jacobians in LLMs would be computationally intractable due to their size. One key technical contribution is thus finding an efficient way of computing Jacobians in this setup. We find that JSAEs extract a relatively large degree of computational sparsity while preserving downstream LLM performance approximately as well as traditional SAEs. We also show that Jacobians are a reasonable proxy for computational sparsity because MLPs are approximately linear when rewritten in the JSAE basis. Lastly, we show that JSAEs achieve a greater degree of computational sparsity on pre-trained LLMs than on the equivalent randomized LLM. This shows that the sparsity of the computational graph appears to be a property that LLMs learn through training, and suggests that JSAEs might be more suitable for understanding learned transformer computations than standard SAEs.

cs.LG↗

arXiv2025

Modeling Nonlinear Oscillator Networks Using Physics-Informed Hybrid Reservoir Computing

Surrogate modeling of non-linear oscillator networks remains challenging due to discrepancies between simplified analytical models and real-world complexity. To bridge this gap, we investigate hybrid reservoir computing, combining reservoir computing with "expert" analytical models. Simulating the absence of an exact model, we first test the surrogate models with parameter errors in their expert model. Second, in a residual physics task, we assess their performance when their expert model lacks key non-linear coupling terms present in an extended ground-truth model. We focus on short-term forecasting across diverse dynamical regimes, evaluating the use of these surrogates for control applications. We show that hybrid reservoir computers generally outperform standard reservoir computers and exhibit greater robustness to parameter tuning. This advantage is less pronounced in the residual physics task. Notably, unlike standard reservoir computers, the performance of the hybrid does not degrade when crossing an observed spectral radius threshold. Furthermore, there is good performance for dynamical regimes not accessible to the expert model, demonstrating the contribution of the reservoir.

eess.SY↗

arXiv2025

Cooperation guides evolution in a minimal model of biological evolution

A challenging simulation of evolutionary dynamics based on a three-state cellular automaton is used as a test of how cooperation can drive the evolution of complex traits. Building on the approach of Wolfram (2025), the fitness of a cellular automaton is determined by the number of updates it endures before returning to an empty configuration. In our population-based model, groups of automata are assessed either by summing individual fitness or by the maximum fitness in the group; this second scheme mimics cooperation by allowing fit rules to support their less fit group mates. We show that cooperation substantially improves the population's ability to evolve longer-lived rules, illustrating a Baldwin Effect for Cooperation: higher diversity across individuals unlock otherwise inaccessible evolutionary paths.

q-bio.PE↗

arXiv2025

Residual Stream Analysis with Multi-Layer SAEs

Sparse autoencoders (SAEs) are a promising approach to interpreting the internal representations of transformer language models. However, SAEs are usually trained separately on each transformer layer, making it difficult to use them to study how information flows across layers. To solve this problem, we introduce the multi-layer SAE (MLSAE): a single SAE trained on the residual stream activation vectors from every transformer layer. Given that the residual stream is understood to preserve information across layers, we expected MLSAE latents to 'switch on' at a token position and remain active at later layers. Interestingly, we find that individual latents are often active at a single layer for a given token or prompt, but the layer at which an individual latent is active may differ for different tokens or prompts. We quantify these phenomena by defining a distribution over layers and considering its variance. We find that the variance of the distributions of latent activations over layers is about two orders of magnitude greater when aggregating over tokens compared with a single token. For larger underlying models, the degree to which latents are active at multiple layers increases, which is consistent with the fact that the residual stream activation vectors at adjacent layers become more similar. Finally, we relax the assumption that the residual stream basis is the same at every layer by applying pre-trained tuned-lens transformations, but our findings remain qualitatively similar. Our results represent a new approach to understanding how representations change as they flow through transformers. We release our code to train and analyze MLSAEs at https://github.com/tim-lawson/mlsae.

cs.LG↗

arXiv2024

An iterated learning model of language change that mixes supervised and unsupervised learning

The iterated learning model is an agent model which simulates the transmission of of language from generation to generation. It is used to study how the language adapts to pressures imposed by transmission. In each iteration, a language tutor exposes a naïve pupil to a limited training set of utterances, each pairing a random meaning with the signal that conveys it. Then the pupil becomes a tutor for a new naïve pupil in the next iteration. The transmission bottleneck ensures that tutors must generalize beyond the training set that they experienced. Repeated cycles of learning and generalization can result in a language that is expressive, compositional and stable. Previously, the agents in the iterated learning model mapped signals to meanings using an artificial neural network but relied on an unrealistic and computationally expensive process of obversion to map meanings to signals. Here, both maps are neural networks, trained separately through supervised learning and together through unsupervised learning in the form of an autoencoder. This avoids the computational burden entailed in obversion and introduces a mixture of supervised and unsupervised learning as observed during language learning in children. The new model demonstrates a linear relationship between the dimensionality of meaning-signal space and effective bottleneck size and suggests that internal reflection on potential utterances is important in language learning and evolution.

cs.CL↗

arXiv2024

Cooperation as well as learning: A commentary on 'How learning can guide evolution' by Hinton and Nowlan

According to the Baldwin Effect learning can guide evolution. This does not suppose that information about what has been learned is transferred back into the genetic code: in the Baldwin Effect complex multi-gene characteristics are discovered through learning but acquired through standard selectionist evolutionary processes. Learning serves to improve the search by giving value to a partial, and otherwise useless, subset of the required genes. An elegant and concrete treatment of the Baldwin Effect is given in 'How learning can guide evolution', a 1987 paper by G. E. Hinton and S. J. Nowlan. This includes a simple, but revealing, simulation illustrating the effect. As a commentary on that paper, a similar simulation is used here to demonstrated that cooperation can also guide evolution. Like learning, cooperation has a clear benefit to survival, but what is proposed here is small addition: that cooperation, like learning in the Baldwin Effect, can also allow complex characteristic to be discovered and acquired much faster than they otherwise would. This suggests an additional benefit of social behavior and suggests that social animals have a broader evolutionary path towards some complex adaptations.

q-bio.PE↗

arXiv2024

Modeling language contact with the Iterated Learning Model

Contact between languages has the potential to transmit vocabulary and other language features; however, this does not always happen. Here, an iterated learning model is used to examine, in a simple way, the resistance of languages to change during language contact. Iterated learning models are agent-based models of language change, they demonstrate that languages that are expressive and compositional arise spontaneously as a consequence of a language transmission bottleneck. A recently introduced type of iterated learning model, the Semi-Supervised ILM is used to simulate language contact. These simulations do not include many of the complex factors involved in language contact and do not model a population of speakers; nonetheless the model demonstrates that the dynamics which lead languages in the model to spontaneously become expressive and compositional, also cause a language to maintain its core traits even after mixing with another language.

cs.CL↗

arXiv2024

Investigating the Timescales of Language Processing with EEG and Language Models

This study explores the temporal dynamics of language processing by examining the alignment between word representations from a pre-trained transformer-based language model, and EEG data. Using a Temporal Response Function (TRF) model, we investigate how neural activity corresponds to model representations across different layers, revealing insights into the interaction between artificial language models and brain responses during language comprehension. Our analysis reveals patterns in TRFs from distinct layers, highlighting varying contributions to lexical and compositional processing. Additionally, we used linear discriminant analysis (LDA) to isolate part-of-speech (POS) representations, offering insights into their influence on neural responses and the underlying mechanisms of syntactic processing. These findings underscore EEG's utility for probing language processing dynamics with high temporal resolution. By bridging artificial language models and neural activity, this study advances our understanding of their interaction at fine timescales.

cs.CL↗

arXiv2024

Signatures of Bayesian inference emerge from energy efficient synapses

Biological synaptic transmission is unreliable, and this unreliability likely degrades neural circuit performance. While there are biophysical mechanisms that can increase reliability, for instance by increasing vesicle release probability, these mechanisms cost energy. We examined four such mechanisms along with the associated scaling of the energetic costs. We then embedded these energetic costs for reliability in artificial neural networks (ANN) with trainable stochastic synapses, and trained these networks on standard image classification tasks. The resulting networks revealed a tradeoff between circuit performance and the energetic cost of synaptic reliability. Additionally, the optimised networks exhibited two testable predictions consistent with pre-existing experimental data. Specifically, synapses with lower variability tended to have 1) higher input firing rates and 2) lower learning rates. Surprisingly, these predictions also arise when synapse statistics are inferred through Bayesian inference. Indeed, we were able to find a formal, theoretical link between the performance-reliability cost tradeoff and Bayesian inference. This connection suggests two incompatible possibilities: evolution may have chanced upon a scheme for implementing Bayesian inference by optimising energy efficiency, or alternatively, energy efficient synapses may display signatures of Bayesian inference without actually using Bayes to reason about uncertainty.

q-bio.NC↗

arXiv2024

Nearest-Neighbours Estimators for Conditional Mutual Information

The conditional mutual information quantifies the conditional dependence of two random variables. It has numerous applications; it forms, for example, part of the definition of transfer entropy, a common measure of the causal relationship between time series. It does, however, require a lot of data to estimate accurately and suffers the curse of dimensionality, limiting its application in machine learning and data science. However, the Kozachenko-Leonenko approach can address this problem: it is possible, in this approach to define a nearest-neighbour estimator which depends only on the distance between data points and not on the dimension of the data. Furthermore, the bias can be calculated analytically for this estimator. Here this estimator is described and is tested on simulated data.

cs.IT↗

arXiv2024

Investigating grammatical abstraction in language models using few-shot learning of novel noun gender

Humans can learn a new word and infer its grammatical properties from very few examples. They have an abstract notion of linguistic properties like grammatical gender and agreement rules that can be applied to novel syntactic contexts and words. Drawing inspiration from psycholinguistics, we conduct a noun learning experiment to assess whether an LSTM and a decoder-only transformer can achieve human-like abstraction of grammatical gender in French. Language models were tasked with learning the gender of a novel noun embedding from a few examples in one grammatical agreement context and predicting agreement in another, unseen context. We find that both language models effectively generalise novel noun gender from one to two learning examples and apply the learnt gender across agreement contexts, albeit with a bias for the masculine gender category. Importantly, the few-shot updates were only applied to the embedding layers, demonstrating that models encode sufficient gender information within the word embedding space. While the generalisation behaviour of models suggests that they represent grammatical gender as an abstract category, like humans, further work is needed to explore the details of how exactly this is implemented. For a comparative perspective with human behaviour, we conducted an analogous one-shot novel noun gender learning experiment, which revealed that native French speakers, like language models, also exhibited a masculine gender bias and are not excellent one-shot learners either.

cs.CL↗

arXiv2023

Estimating mutual information for spike trains: a bird song example

Zebra finch are a model animal used in the study of audition. They are adept at recognizing zebra finch songs and the neural pathway involved in song recognition is well studied. Here, this example is used to illustrate the estimation of mutual information between stimulus and response using a Kozachenko-Leonenko estimator. The challenge in calculating mutual information for spike trains is that there are no obvious coordinates for the data. The Kozachenko-Leonenko estimator does not require coordinates, it relies only on the distance between data points. In the case of bird song, estimating the mutual information demonstrates that the information content of spiking does not diminish as the song progresses.

cs.IT↗