Dave Hilbert Retirement Conference

On Sept 4th and 5th 2026, the UIC philosophy department will host a conference in honor of the retirement of David Hilbert. The conference will celebrate Dave's career by hosting talks which engage with his work on color and on George Berkeley. The program is below. 

All are encouraged to attend. Attendance is free but registration is required. Please register by emailing hilbert.conference@gmail.com. Any questions about the conference can also be addressed there. 

Program

Sept 4. 

Location: UIC Institute for the Humanities

Behavioral Sciences Building, 

1007 W Harrison Street, Suite 153

10:30am-12pm: Will Davies (Oxford)

Title and Abstract TBD

12-1:00pm: Lunch

1-2:30pm: Lisa Downing (OSU)

Title and Abstract TBD

2:30-2:45pm: Break

2:45-4:15pm: Jonathan Cohen (UCSD)

Color Constancy, Complexity, and Extrapolationism (or: What Twenty Years of Color Science Teaches Us About Constancy) 

Hilbert's work on color constancy has been central to the philosophy of color for decades; this talk revisits constancy in light of more recent research. A lesson of vision science over the last twenty years is that there is substantial disunity and complexity in "the" phenomenon of color constancy. While a few have flagged this complexity, it has had limited uptake in philosophical discussions, which still tend to treat constancy as a single, relatively homogeneous phenomenon. Appreciating the complexity of constancy might well lead to skepticism --- e.g., the view that "color constancy" picks out something real but irreducibly heterogenous, or that it fails to refer at all. I'll argue that these are overreactions, and will offer a unified computational account of color constancy, drawing on and revising a view I have defended elsewhere, that makes room for (and, indeed, predicts) the observed heterogeneity. 

Sept 5. 

Location: ARC Building, room 241

940 W Harrison St 

10:00-11.30am: Colin Klein (ANU)

Diffusion models represent directively (joint work w. Nuhu Osman Attah)  

Diffusion models lie behind most sophisticated image-generating AI. How they work raises a number of philosophical questions. Notably: it doesn’t seem like they represent facts about the pictures that they generate, or the categories of items they depict, or really anything at all other than statistical facts about the distribution of noise. We argue that diffusion models represent, but they represent directively rather than descriptively. We suggest that this might make for an interesting architectural taxonomy of AI systems, and raises interesting questions for the project of mechanistic interpretability. 

11.30am-11.45pm: Break

11:45am-1.15pm: Dave Hilbert (UIC)

Title and Abstract TBD

Organizers: Rachel Goodman and Aidan Gray

The conference is sponsored by the UIC philosophy department and by the UIC Institute for the Humanities.