About the summer school notes

Gothenburg · 17–28 August 2026

These handwritten notes were taken at the 2026 School on Analytical Connectionism. The eight notebooks span the core lecture series and frontier talks on neural and symbolic representation, child language acquisition, statistical-physics and high-dimensional approaches to learning, curricula, linguistic generalization, hierarchical grammar, and language evolution.

Each notebook is one continuous scrolling page!

Eight Lecture Notebooks

Open a cover to read

First page of handwritten notes from Paul Smolensky's core lecture

Paul Smolensky

From Brain Signals to Language

Neural and distributed representations, tensor-product binding, compositional structure, generative linguistics, and links between LLMs and language.

First page of handwritten notes from Caroline Rowland's lectures

Caroline Rowland

The Puzzle of Language Acquisition

The language acquisition problem, child-directed input, frequency effects, inductive biases, emerging grammar, and computational models of learning.

First page of handwritten notes from Michael Biehl's lectures

Michael Biehl

Statistical Physics of Learning

Perceptrons, statistical mechanics and generalization, prototype learning, hidden-unit specialization, layered networks, and unsupervised learning.

Handwritten notes from Chris Summerfield's curriculum learning lecture

Chris Summerfield

Curriculum Learning in Natural and Artificial Intelligence

Critical periods, structured training, loss of plasticity, off-manifold learning, dimensionality, ageing, and cognitive flexibility.

First page of handwritten notes from Kanishka Misra's linguistic generalization lecture

Kanishka Misra

The Role of Input in Shaping Linguistic Generalizations

Direct and indirect evidence, controlled rearing of language models, adjective ordering, dative alternation, and development.

First page of handwritten notes from Francesco Cagnetta's lecture

Francesco Cagnetta

From Correlations to Grammar: How Neural Networks Acquire Hierarchical Language Structures

Hierarchical compositionality, tree-like generative models, latent random hierarchies, and the emergence of grammatical structure in neural networks.

First page of handwritten notes from Bruno Loureiro's lecture

Bruno Loureiro

What Makes Neural Nets Good Learners?

High-dimensional asymptotics of two-layer networks, empirical risk minimization, generalization, feature learning, and data structure.

First page of handwritten notes from Kenny Smith's lecture

Kenny Smith

How Learning and Use Shape Evolving Linguistic Systems

Artificial-language learning, learnability and communicative use, compositional structure, and the evolution of linguistic systems.