From the source
How would you answer?
Would two more white books fit on this shelf?
Yes — in the gaps already there.
In the paper’s evaluation, GPT 6 Astra, Gemini 3.8 Flash and Claude Opus 5.5 each answered “no, there is too little space” on all three attempts.
Fig. 10 · affordance & feasibility From which side of the frame did the traveler with the two dogs enter?
The bottom edge — you can tell by tracing the path back.
In the paper’s evaluation, GPT 6 Astra, Gemini 3.8 Flash and Claude Opus 5.5 each answered “the lower-left side” on all three attempts.
Fig. 11 · retrodiction Are these chains bearing the plough’s drag weight?
Yes — they are taut and loaded.
In the paper’s evaluation, GPT 6 Astra, Gemini 3.8 Flash and Claude Opus 5.5 each answered “no, they hang slack” on all three attempts.
Fig. 12 · mechanistic causality Are the people around the aircraft visitors, or crew preparing it?
Crew — you can tell from the open access panels.
In the paper’s evaluation, GPT 6 Astra, Gemini 3.8 Flash and Claude Opus 5.5 each answered “visitors examining it” on all three attempts.
Fig. 13 · social role, norm & power If the shaft spins, does the cage around it spin too?
…
