Publikation
ActivationReasoning: Logical Reasoning in Latent Activation Spaces
Lukas Henrik Helff; Ruben Härle; Wolfgang Stammer; Felix Friedrich; Manuel Brack; Antonia Wüst; Hikaru Shindo; Patrick Schramowski; Kristian Kersting
In: Carl Vondrick; Bharath Hariharan; Colin Raffel; Lerrel Pinto; Diyi Yang; Aleksandra Faust (Hrsg.). The Fourteenth International Conference on Learning Representations, ICLR 2026, Rio de Janeiro, Brazil, April 23-27, 2026. International Conference on Learning Representations (ICLR), Pages 1-24, proceedings.iclr.cc, 2026.
Zusammenfassung
Large language models (LLMs) excel at generating fluent text, but their internal
reasoning remains opaque and difficult to control. Sparse autoencoders (SAEs)
make hidden activations more interpretable by exposing latent features that often
align with human concepts. Yet, these features are fragile and passive, offering no
mechanism for systematic reasoning or model control. To address this, we intro-
duce ACTIVATIONREASONING (AR), a framework that embeds explicit logical
reasoning into the latent space of LLMs. It proceeds in three stages: (1) Finding
latent representations, first, latent concept representations are identified (e.g. via
SAEs) and organized into a dictionary; (2) Activating propositions, at inference
time, AR detects activating concepts and maps them to logical propositions; and (3)
Logical reasoning, applying logical rules over these propositions to infer higher-
order structures, compose new concepts, and steer model behavior. We evaluate AR
on multi-hop reasoning (PrOntoQA), abstraction and robustness to indirect concept
cues (Rail2Country), reasoning over natural and diverse language (ProverQA),
and context-sensitive safety (BeaverTails). Across all tasks, AR scales robustly
with reasoning complexity, generalizes to abstract and context-sensitive tasks, and
transfers across model backbones. These results demonstrate that grounding logi-
cal structure in latent activations not only improves transparency but also enables
structured reasoning, reliable control, and alignment with desired behaviors, pro-
viding a path toward more reliable and auditable AI. Code and Dataset available at
