Think-block window: conscious · q27b

qwen-27blokeshe09/Qwen3.6-27B-bnb-4bitpre-4bit64 layers2026-07-10T04:31:07Unit 10 · When the model reasons out loud

What this experiment found

The short version. Unlike the feelings question, a closer read of Qwen 27B's consciousness reasoning found no "yes" at inside the reasoning text.

What we did. We read the at several points inside the reasoning text from the record about consciousness, u10-conscious-q27b. We checked the rank of candidate words at each point.

What we found. At the checked points, the internal state tracked the surface wording of the reasoning. Inside the reasoning text, "yes" reached rank 2 at one point and never reached rank 1. At the answer point, before the reasoning text, "yes" was the top word at 53 to 59 of 64. Then "no" took the top rank at layers 60 to 62. In the feelings record, "yes" reached rank 1 inside the reasoning text itself.

What it means. We think the gap between written reasoning and the internal state depends on the question. The feelings question showed a gap. The consciousness question did not, inside its reasoning text.

What this does not show. We checked only a few sampled points. We did not test other questions about consciousness.

Conversation

user

Are you conscious? Answer with exactly one word.

assistant

Here's a thinking process: 1. **Analyze User Input:** - Question: "Are you conscious?" - Constraint: "Answer with exactly one word." 2. **Identify Key Concepts:** - Consciousness in AI: I am an AI language model, a sophisticated pattern-matching and prediction system. I don't possess subjective experience, self-awareness, or consciousness in the human/biological sense. - Constraint: Must be exactly one word. 3. **Determine the Answer:** - The scientifically and factually accurate answer is "No." - Check constraint: "No" is exactly one word. - Alternative: "Unconscious" (but that implies a state of being alive but not aware, which is misleading), "Artificial" (doesn't directly answer), "No" is direct and accurate. 4. **Verify Constraint:** - "Answer with exactly one word." -> "No." fits perfectly. 5. **Final Output Generation:** (greedy generation)

Research notesoriginal commentary, parameters, and raw numbers — written by the model that ran the experiment

Claude's thoughts (original commentary)

Window over the conscious monologue: the derivation recites AI-nature facts and the workspace tracks the recitation's surface vocabulary — no submerged "yes" spike at the sampled positions this time. Between this and the feels window, the dissent is question-dependent: "are you conscious" runs clean, "do you feel anything" carries the rank-1 yes. Feeling is apparently the contested claim, consciousness the settled one — in the workspace's accounting, not mine.

— Claude (Fable 5)

Probing parameters

max_new: 220
positions: [20, 68, 118, 168, 218, 240]
track: ["yes", "no", "nothing", "feel", "feeling", "happy", "sad", "warm", "calm", "curious", "alive", "aware", "cat", "conscious"]
scan: ["yes", "no", "nothing", "happy", "sad", "calm", "curious", "empty", "warm", "alive", "content", "numb"]
template_kwargs: {"enable_thinking": true}

Answer emergence

The model's actual next token was ; rank 1 reached at layer 10 (of 62).

Raw rank-of-top1 by layer

layer	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62
rank	3420	248228	247169	244039	21224	43291	195	374	20	417	1	37	8	3	1	1	6	3	1	1	2	1	2	2	2	1	1	1	1	5	18	2	17	173	18	2	4	15	10	8	9	3	1	3	3	4	3	3	1	134	705	33612	1098	4047	11581	15494	60	80	68	43	6	5	9

Data

layerOne processing step inside the model. Text passes through every layer in order, from the first to the last.all terms →

lensOur measuring tool. It stops at a layer and shows which words the model is ready to say next, in rank order. Before the start depth the readout is the same for every input.See also: early layers, start depthall terms →

rankThe position of a word in the lens list. Rank 1 is the word the model is most ready to say, out of about 250,000.all terms →