Source
Bechara, Damasio, Tranel & Damasio (1997), Deciding Advantageously Before Knowing the Advantageous Strategy, LIT-383 (Science 275:1293–1295), read in full in NOTE-331 with its notes.
What was actually shown
Design. Ten normal controls and six patients with bilateral ventromedial prefrontal lesions and real-life decision defects. Four decks: A and B pay 50 with small penalties and win. Play stopped after 100 cards. From card 20, and every 10 cards after, play was interrupted to ask what the participant knew and felt about the game. The answers define periods: pre-punishment (before the first loss, about card 10), pre-hunch (“no notion of what was happening”; controls roughly cards 10–50), hunch (liking or disliking decks without certainty; controls roughly 50–80; never reached by patients), and conceptual (can say which decks are good and why; reached by 7 of 10 controls and 3 of 6 patients).
Anticipatory SCRs (note 8). All controls and no patients generated them: Fisher’s exact test, one-sided P < 0.001. In a group × period × deck ANOVA over the pre-punishment and pre-hunch periods, the reported key effect is group × period, F(1,14) = 16.24, P < 0.001: SCRs rose from baseline to pre-hunch only in controls. No deck effect or deck interaction is reported for those periods. In the hunch and conceptual periods, bad-deck SCRs were “generally higher” than good-deck ones, and the difference “did not reach statistical significance”.
Choices (note 9). Group × period × deck, F(1,14) = 6.9, P < 0.02. Pre-hunch: patients chose significantly more bad than good cards (P < 0.01); controls chose more good than bad, “not statistically significant”. Hunch and conceptual: controls chose significantly more good cards (P < 0.01). The three patients who reached the conceptual period still chose more bad cards, not significantly; so “still chose badly” means they did not switch.
What could have come out otherwise. Some patients could have developed anticipatory SCRs, or the patients who could state the strategy could have followed it. Neither happened. The controls’ pre-hunch preference could have been significant, and was not.
What this does not say
- That nonconscious biases guide behaviour before conscious knowledge. That is the paper’s inference (C5 in NOTE-331), and it rests on its two weakest results. The preference became significant only when controls already reported liking and disliking decks. “Before the articulated strategy” is supported; “before any awareness” is not.
- That controls responded to the risky decks before knowing they were risky. The pre-hunch test reported is group × period, not bad versus good deck, and later deck differences are not significant either.
- That the awareness probe was fine-grained. Two open questions, from card 20, every 10 cards. Knowledge not volunteered is coded as absent, and nothing is asked in cards 10–20, when the SCRs are said to begin.
- That the SCR causes the choice. The two are measured together.
- That the somatic marker hypothesis is confirmed. The two-chain model and the ventromedial role in it (p. 1294, Fig. 2) are stated, not tested.
- That the result generalizes beyond small samples. Six patients, three in the conceptual-period result, and control demographics unstated.
Connections
- THEORY-500 states the somatic marker hypothesis this study was offered for, with LIT-383 as a second source. The caveats above bound what that account can take from this experiment: its covert-bias clause is not shown here.
- The hypothesis it is cited for. Damasio’s 1996 statement of the somatic marker hypothesis (LIT-384, NOTE-332) previews this result in a sentence (“long before they have … any notion whatsoever”); this report is the evidence for that sentence and is weaker than it.
- Felt affect. THEORY-045 holds that felt affect is the experience of the body’s regulatory state. The hunch period, liking and disliking decks without knowing why, is the kind of case it would describe; this study measured the SCR, not what was felt.
- The awareness critique. Maia & McClelland (2004), which re-examined what participants know in this task, is not held in either record and was not read.
- Censors. NOTE-361 pairs this result with Minsky’s learned censors (LIT-408) as a case against a clean affect–intellect split. That pairing is the reader’s, and neither work cites the other.