SSP Forum: Fisher Anderson and Yiheng Yao (M.S. Candidates)

The
Symbolic Systems Forum
(community sessions of SYMSYS 280 - Symbolic Systems Research Seminar)
presents

Which Levels of Human Consciousness Pose a Scientific Question?

Fisher Anderson (M.S. Candidate)
Symbolic Systems Program

and

Talk is Cheap, Communication is Hard: Dynamic Grounding Failures and Repair in Multi-Agent Negotiation

Yiheng Yao (M.S. Candidate)
Symbolic Systems Program

Monday, April 20, 2026
12;30-1:20 pm PT
Computing and Data Science Building (CoDa), Room E160
In-person event, not recorded

Note: Lunch is provided, if pre-ordered, only for members of SYMSYS 280, but others are welcome to bring a lunch and eat during the presentation.

Abstracts:

Fisher Anderson, "Which Levels of Human Consciousness Pose a Scientific Question?" (primary advisor: Chris Potts, Linguistics)
     The word "consciousness" is one of the most frequently used and least consistently defined terms in both scientific and ordinary discourse. This talk analyzes a five-level taxonomy of human consciousness,ranging from simple awareness to subjective experience, and asks a targeted question of each level: does it meet the basic criterion of scientific inquiry – can it be falsified? The answer falls broadly into three distinct categories, with the final proving surprisingly impervious to falsification. The result is not skepticism about consciousness, but a principled demarcation whose stakes extend well beyond the classroom.

Yiheng Yao, "Talk is Cheap, Communication is Hard: Dynamic Grounding Failures and Repair in Multi-Agent Negotiation" (primary advisor: Tobias Gerstenberg, Psychology)
     Grounding is the collaborative process of establishing mutual belief sufficient for the current communicative purpose. While static grounding maps language to a shared, externally observable context, dynamic grounding is a joint activity where meaning is negotiated through interaction. Current multi-agent Large Language Model (LLM) benchmarks focus on static, one-shot tasks, overlooking the ability to repair grounding breakdowns across turns. We introduce an iterated, multi-turn negotiation game in which two agents allocate shared resources toward private projects with verifiable jointly optimal outcomes. While individual agents can identify optimal allocations in isolation, agent dyads consistently fail to reach them across open- and closed-source models. Our investigation reveals four failure modes: (1) coordination degrades more in shifting dyads lacking shared interaction history; (2) a reliance on perfunctory fairness (equal resource splits) over reward-maximizing coordination; (3) a stubborn anchoring effect where initial proposals are treated as axiomatic rather than negotiable; and (4) failures in referential binding, where agents lose track of commitments across turns. These results highlight dynamic grounding as a critical and understudied axis of multi-agent coordination. Our framework isolates communication failures and enables systematic study of interventions including structured communication scaffolding and theory-of-mind elicitation. (Joint work with Robert Hawkins)