Psychology, language, and computational models
Research questions, selected publications, and software associated with my work.
Conceptual representations and semantic geometry
My research examines what a computational representation should preserve about a concept. Single vectors are useful summaries, but may lose important information about contextual variation, relations between concepts, and representational change. Semantic subspaces offer one possible mathematical alternative; their advantages must be established empirically rather than assumed.
This work includes Quantum projections on conceptual subspaces (2023) and a methodological discussion published in Science Progress (2024). My provisional research programme, Beyond Point Representation, develops related questions about representational structure and psychological evidence.
My PhD thesis, De vectores a conceptos: nuevas rutas en la representación computacional del lenguaje (Autonomous University of Madrid, 2024), provides the earlier framework for this line of work. It was supervised by Guillermo Jorge-Botana and Ricardo Olmos. Thesis record in Dialnet ↗ · Full text in the UAM repository ↗.
Language models and psychological measurement
I study whether measures derived from computational language models capture psychological constructs, with particular attention to construct validity, out-of-sample generalization, and competing operationalizations. In 2026 I co-authored a study of language-model surprisal as an individual-difference measure.
Our cross-validation study of LIWC linguistic indicators (2022) further demonstrates why exploratory associations should be evaluated beyond their original samples.
Quantum-inspired cognitive modeling
I use tools from quantum-inspired probability and geometry as formal descriptions of selected cognitive phenomena. This does not imply that the brain is literally a quantum computer. The relevant question is whether the models yield falsifiable predictions and explain behavioral patterns beyond simpler classical alternatives.
Code and research tools
Selected public repositories associated with my research and teaching. Some are research prototypes rather than maintained or fully documented software releases.
QLang
Tools for extracting contextual language examples, building semantic representations, and analyzing concept subspaces.
QLSA
Prototype methods for constructing conceptual subspaces in LSA semantic spaces. The repository documents access dependencies for the original semantic resource.
AGRUPA
Utilities for normalizing artwork descriptions and computing dictionary-based psychological measures.
SampleSizeApp
Sample size calculations for different experimental and statistical designs.
LDAapp Cualitativas
Teaching tool to compare quantitative and qualitative analysis of interview transcripts.
Collaborations and research supervision
I contribute to AGRUPA (PID2024-157794NB-I00), a UAM-led research project examining stereotypes in cultural artifacts, as a member of its working team. I am currently based at UNED, where I teach statistics, computational methods, and research methodology in Psychology.
For research collaborations, student projects, or doctoral inquiries, see the institutional contact details on my homepage.
Teaching statistics and programming is closely connected to my methodological work. Courses and official guides.