Philosophy of Psychology 108 1 Blocks Methodological Puzzle
心理學哲學 Philosophy of Psychology 108 -1 第六週
Block’s Methodological Puzzle �“[S]ince report is how one gain access to consciousness and report implicates attention, then it looks like the primary evidence for consciousness cannot also provide evidence for consciousness in the absence of attention” (p. 187). �我們似乎只能透過主體的報告來知道他是否有意識, 但報告蘊涵了注意力,這樣我們要如何把注意力跟意 識在實驗上區分開?
Inference to the Best Explanation � 1. Visual working memory (the workspace) has a limited capacity. � 2. Overflow: Phenomenology has a higher capacity than working memory. � 3. “The control of working memory is in the front of the head” (Block, 2007, p. 496). � 4. Arguably, the “core neural basis of visual phenomenology is in the back of the head” (ibid. ). � 5. If one assumes that the machinery controlling working memory is necessary for visual phenomenology, then one cannot explain overflow. � 6. If one assumes that the machinery controlling working memory is not necessary for visual phenomenology, then one can explain overflow.
Sperling’s Partial Report Paradigm �“The information available in brief visual presentations” (Sperling, 1960). �Research question: “How much does one see in a glance” (p. 187)? �“Subjects typically claimed to see more than they could report” (p. 187). �https: //youtu. be/ACddnsfg. J 7 I
A Further Question �“Does one see more than can be remembered” (p. 188)?
Summing Reports �“[B]y taking the number of letters reported in partial reports and multiplying by the number of lines in the array, he could obtain an estimate of the total number of letters that were seen” (p. 189). �“Doing this, Sperling found the number of letters reported to be on average 9. 1, about three of four letters in each line” (p. 189). �如此下結論說受試者平均有看到 9個字母/數字是否言 之過早?
Other Details �“[V]isual capacity exceeds working memory capacity” (p. 189). �Partial Report Advantage: 在部分報導的情況下,受試 者可以記得更多字母或是數字。 �Stimulus offset: 在刺激物消失的300毫秒為止,受試者 還是可以有如上的PRA。 �Immediate Memory = Working Memory
Visual Persistence �Max Coltheart (1980) � Neural Persistence: “the persistent activity of visual neurons after the stimulus is removed. ” � Visible Persistence: “the continued visibility of the stimulus after offset, such as in an afterimage. ” � Informational Persistence: “the continued accessibility of the stimulus after offset” (p. 189). �Ulric Neisser (1967): Informational Persistence = Iconic Memory. Cf. Echoic memory.
From Information to Consciousness �“The crucial next question is whether iconic memory, what Sperling uncovered in his partial report paradigm, reflects conscious or unconscious perception” (p. 189). �“If it reflects conscious perception, then the capacity of phenomenology in iconic memory exceeds the capacity of working memory” (p. 189). �For Overflow, i. e. , against Gatekeeper.
Contents of Visual Memories � 表徵內容 Representational Contents: � 語句/思想:具有真值條件(truth conditions)。 � 圖像/經驗:具有準確條件(accuracy conditions)。 � Kinds of Contents: � Unconscious: The information is specific and unconscious (or reflects unconsciousness), but can be brought to consciousness, say by attention. � Nonspecific: The information is conscious (or reflects consciousness), but in some way it is nonspecific, though it can be rendered more specific due to attention. � Specific: : The information is conscious (or reflects consciousness) and highly specific. (p. 190)
Block’s Crucial Distinction (1) Generic and Specific Phenomenology: “For the Sperling experiment, the relevant generic/specific difference would be that between a phenomenal presentation that there is an array of alphanumeric characters and a phenomenal presentation of specific shapes of all or most items in the array…My argument was that before the cue, there is specific phenomenology for all or almost all items. ” (Block, 2007, p. 531; original italics)
Block’s Crucial Distinction (2) � A. Retinal-Based Visual Iconic Memory: Block also calls this ‘pure visual iconic memory’ (2011: 567). � B. Fragile Visual Short-Term Memory (fragile VSTM): It is a ‘intermediate’ form of visual memory (ibid. ) and ‘can last 4 -5 seconds. ’ � C. Visual Working Memory: It ‘can persist much longer and has a smaller capacity than any form of iconic memory and probably requires categorization’ (ibid. ).
Block’s View Introduced �Phenomenal consciousness overflows (i. e. , is richer than) cognitive access (≠ accessibility, accessconsciousness). �More specifically, the claim is that fragile VSTM is conscious and it has a larger capacity than visual working memory. �The key in the view is that before the cue, we have specific phenomenology for almost every item, and those phenomenology is specific enough for reports.
Reporting Phenomenology �Specific: “the information regarding each letter identity in memory is sufficient to support report of the identity of each letter when appropriately cued (Sperling’s result)” (p. 190). �Wayne Wu’s Phenomenology: “Yet it also seems that the letters I cannot report (the unreported letters) are nevertheless visible to me. I see something as those positions although they don’t appear to be specific letters. Rather, the unreported letters seem to be a smudge, as if blurrily seen, perhaps not even symbollike” (p. 191).
Reports and Phenomenology �Another perspective: “subjects do have access to all the letters, but possibly in different degrees (see also Stazicker 2011, section 3)” (p. 191). �作者認為Overflow必須跟Specific結合,因為 Nonspecific跟Gatekeeper是相容的。這個判定是正確 的嗎? �在很大的程度上理論者都還是很依賴自己的內省;這 樣的方法是該被容許的嗎?Cf. Kouider et al. , “fragmentary phenomenology” (2010).
Block on Specificity � Measuring capacity: 4 letter identity + ? � Block’s reasons for specificity: � 1. Subjects in experiments attest to drawing on specific phenomenology in making their partial reports. � 2. Denying specific phenomenology suggests that, when subjects have specific phenomenology restricted to the specific letters they report, then there is a shift from unconsciousness or generic phenomenology to specific phenomenology. Subjects should notice a change, but they do not. � 但在時間那麼短的狀況下,受試者沒有注意到似乎也是可 能的。
Change Blindness Again �Block認為,如果一開始受試者不是對多數字母/數字 都有specific phenomenology,那經驗上的改變應該會 讓他們發現。 �但考慮Change blindness: “even when a subject focuses attention, they miss substantial changes in visual scene” (p. 193). �Inference to the Best Explanation: 在遇到類似質疑時, Block常常說他的立場是透過abduction而來;但問題 是我們如何決定哪一個理論是best?
Summing Reports, Again �前面提到Sperling認為把partial reports的結果加總, 可以說受試者大約有看到 9個字母。 �“It would be appropriate if the iconic memory representation is unaffected by any further processing induced by the cue” (p. 194). �“In particular, the representation must not be affected by attention as induced by the cue” (p. 194).
Postdiction �The postdiction interpretation has been argued by Ian Phillips (2011). The gist of the view is that attentional effects retrospectively modulate earlier stimuli. �A paradigmatic case: Backward Visual Masking � 50 ~ 100 Msecs will be most obvious (Alpern 1953)
Summary of Phillips’ Argument �Step 1. Prima Facie Plausibility (Alpern 1953) �Step 2. Longer delay (Weisstein and Wong 1986, Watanabe and Shimojo 2001) �Step 3. Cross-modal effect (Sekuler, Sekuler and Lau 1997)
Stazicker’s View Compared �The main claim: “[Participants’] visual experience should be characterized as an experience as of nine characters arranged in a grid, only three or four of them appearing sufficiently determinately to appear as a specific letter. ” (2011, p. 169) �“[The] effect of cuing…was to alter their conscious experience such that some information became more determinate in it” (ibid. )
Stazicker’s View Compared (Cont. ) �Stazicker then insists that “this interpretation doesn’t require any retroactive effect of attention on conscious experience. ” (ibid. , original italics) �My reply: It is correct to say that ‘it doesn’t require’ retrospective effect, but this is compatible with the view that it actually happens in the Sperling case. Given that the cue comes after the stimuli offset, it seems right to say that in this case retrospective effect does happen.
Weak Overflow Introduced �Both Block and I interpret participants’ reports as saying that they read off answers from their visual phenomenology. �I believe a weaker view can accommodate this evidence: Before the cue, the participants have generic phenomenology, and some parts of it are more specific (fixation, attention, etc. ). We have specific phenomenology for almost every item but only 3 -4 of them are specific enough for reports.
Postdiction and Overflow �Weak overflow seems to be a version of the postdiction interpretation, since it says that after the cue, attention (retrospectively) makes some parts of the overall phenomenology more specific. �But I believe postdiction and weak overflow are compatible, contra Phillips and Block, since even if postdiction does occur in the Sperling case, it is still sensible to say that before the cue, some bits of phenomenology are too rich to be captured by cognitive access at any given moment (as Block argues)
Answer to Block’s Objection �Block thinks the Lamme experiment (2003) shows that postdiction cannot accommodate the relevant effects, since the delay is too long (several seconds for trained subjects). �However, the forced choice method used here overestimate consciousness. �Therefore weak overflow/postdiction is still a possible option.
- Slides: 30