Understanding Transformers and Large Language Models (LLMs) with Natural Language Processing (NLP)

Dear Colleagues,

Large Language Models (LLMs), and transformers in general, have reshaped the scientific landscape of Natural Language Processing; however, their internal reasoning and divergence from human cognition remain largely opaque. This Special Issue, edited by Prof. Massimo Stella and Dr. Alexis Carrillo Ramirez, invites innovative research that leverages NLP as a lens to better understand LLMs.

We invite you to contribute to this Special Issue with novel works related to the following areas of research:

1. Designing linguistic stimuli, psycholinguistic benchmarks, or contrastive prompts to uncover latent representations and emergent abilities of LLMs in NLP tasks, e.g., harnessing LLMs’ ability in detecting human emotions, creativity levels, personality traits, and other elements from language.
2. Comparing distributional, causal, and neuro‑symbolic analyses of transformer layers with human behavioral and neural data, e.g., comparing distributional representations of key linguistic units.
3. Tracing how training corpora, fine‑tuning, and in‑context learning shape model predictions, biases, and/or theory of mind relative to LLMs, e.g., assessing affective or cognitive biases within LLMs.

We particularly encourage interdisciplinary work bridging NLP, computational linguistics, cognitive science, network science, and ethics to illuminate both the potential and the limits of scaling laws. Ultimately, this Special Issue aims to develop principled methodologies that convert black‑box performance into transparent knowledge, guiding the responsible development of future AI systems aligned with human communicative norms. Submissions discussing reproducibility protocols and open‑source toolkits for community validation are especially welcome.

Prof. Dr. Massimo Stella
Dr. Alexis Carrillo Ramirez
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 250 words) can be sent to the Editorial Office for assessment.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. AI is an international peer-reviewed open access monthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 1800 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• bias in LLMs
• human–AI language comparison
• cognitive benchmarks
• LLMs’ trustworthiness