Talk:The Magical Number Seven, Plus or Minus Two

From Wikipedia, the free encyclopedia
Jump to: navigation, search


Remove Farrington link in intro?


Or spot on? In the section: Other cognitive numeric limits - "Cowan also noted a number of other limits of cognition that point to a 'magical number four'" made, for me, an immediate connection to a masterful performance that totally demonstrates the concept: .


Rather looks to me like she wrote that herself. A minor paper with only 7 citations doesn't justify the use of the word 'demonstrated'.Suggested' may be better, but if this has to be included at it should be in some later section. In fact, I 'd say the paper doesn't really warrant a ref at all. If we included every minor fluff and fart with 7 cites on this topic the page would be 25k words.

So... what is this?[edit]

I've read the article over... Exactly what is 7 +- 2? It doesn't say.

I have recently sat a psychology exam in Cognitive and Developmental psychology. I am very intreaged by Miller's Magic Number Seven, and think that it is a very successful theory. —The preceding unsigned comment was added by (talkcontribs) .

I found it very... bleh. Kinda like something I knew and took for granted, and seeing it published makes me think "what the hell?"
Currently the articles does not adequately define the topic, and goes directly into a sceptical section on the urban legend. Whereas neutral articles would introduce the topic of discussion and represent the various points of view first. A good place to start would be to remove any original research NOR and perhaps add a section that describes the theory and its contribution to cognitive theories of memory. Some closely related topics are Short term memory and Working memory. --Comaze 02:58, 10 September 2006 (UTC)
The 7±2 topic itself is only a brief one, the urban legend has grown out of all proportion to the paper that spawned it. George Miller published a paper that pointed out how various measurements produced answers around the number seven. He did not propose a theory as such and subsequent memory research went off in all sorts of directions. Eventually short term memory researchers came up with a model that explained a 7±2 effect for English digits. What contribution did this paper make to cognitive psychology? It probably drives cognitive psychologists up the wall because it has become part of popular culture and they constantly get asked questions about it. Derek farn 10:23, 10 September 2006 (UTC)
I glanced at your the external links you pointed out, but I think you may be giving too much weight to those non-academic sources. The contrary view is that there are thousands of papers, including many peer reviewed, in psychological literature that cite George Miller's paper. I do not deny that the importance of sequencing information into 7±2 chunks has been somewhat overstated, especially in pop psychology seminars and uncritical pedagogy, possibly explaining your concerns that cognitive psychologists are often asked about the importance of 7±2 chunks in memorisation. As I said there are many counter examples, a quick search on Psychinfo revealed a number of results in recent peer reviewed psychological literature, for example, here is a quote published in Psychological Review, "People are limited in the number of items they can consider at one time. That limit—reflecting the capacity of short-term memory and originally estimated at seven plus or minus two items (Miller, 1956)—figures prominently in explaining performance in numerous cognitive tasks (Simon, 1990). Typically it is regarded as a constraint that hinders performance." (Kareev p.280) Reference: Kareev, Yaakov (2005) And Yet the Small-Sample Effect Does Hold: Reply to Juslin and Olsson (2005) and Anderson, Doherty, Berg, and Friedrich (2005). Psychological Review. 112(1):280-285 --Comaze 11:06, 10 September 2006 (UTC)
What is this article supposed to be about? I'm not sure that I know. Is it a history of the impact the 7±2 paper? I would not call the Urban legends section 'skeptical', it simply points out some 'facts' attributed to this paper are not contained in this paper and that subsequent research has lots to say on this topic. Talking of which we now have The Magical number 4.
A survey of the impact of the 7±2 paper on academic work would be useful, but I am not qualified to write it. People seem to like adding in their own favorite quote (eg, watership down), hopefully we can discourage this noise. Derek farn 12:05, 10 September 2006 (UTC)
Miller is probably more authoritative and deserves more space in the article, certainly in psychological discourse, than Sanjiv Sidhu or "Hrair from Watership Down". Additionally the section on urban legends is important as it assists in dispelling some of the mythology surrounding this topic. I cannot really do a survey at the moment because I'm too busy with work due. A third party comment from an expert may help us out here. --Comaze 13:15, 10 September 2006 (UTC)

Possible merge with Chunking (psychology)[edit]

!!Why should it be merged with chunking? This should be a historical article, explaining that the paper was the original presentation of a theory

The chunking article explains the modern day useage and interpretation of a theory that has some derivation from Miller's one. I think they should be kept seperate Christianpunk 10:27, 13 February 2007 (UTC)

I agree. Miller's paper is notable itself and warrants an article, and chunking continues to be developed and needs its own article. I'm going to remove the merge tags. -R. S. Shaw 04:33, 22 March 2007 (UTC)

Not so magical...[edit]

I have a theory that this has much to do with the way information is stored in our brains.

We don't store information in terms of bits and bytes. We store information in terms of items or atomic elements. An item may be a number, a word, a picture, a sound, a feeling, or any single object.

For example, the set A can be treated as a simple item. A description of the set A (such as "A is the set of all prime digits") is another item. An enumeration of the set A: {2, 3, 5, 7} are four items linked together in a group or set.

Take a look at the following diagrams:

Hrair 5.svg Hrair 7.svg Hrair 9.svg

In all of them, there's a central item, which is the main focus of our attention, and there are 6±2 other items, that are related to the central item.

The diagram in the middle has 7 elements. They're neatly ordered in a 3D cross-like figure, just as we would expect a set of 7 neurons to be arranged. The other two diagrams are just diagrams where there are less or more related items.

Of course, when you add too much links, things start to become unmanageable. The diagram with nine items already looks overcrowded. Try adding just one more item: you won't be able to do so without forgetting another item.

That's my theory about the whole Hrair limit. LGM 16:07, 2 August 2006 (UTC)

This (this little article here, including both the theory and the graphics) is a good illustration of one of the things Miller does say, that adding additional information (such as spatial organization and color) increases the channel capacity. The basic point here ("we store information in terms of items or atomic elements") is probably related to what Miller calls "chunking," although he doesn't present that as a theory of how memory works, but only of how to apply and measure stimuli. Jackrepenning (talk) 00:25, 4 March 2008 (UTC)
I would like to propose this bit to be deleted. Could the user first read a bit more about cognitive science and psychology before trying to dazzle others with the profound knowledge? Articles saying more or less the same are plentiful, but tend to be backed up by evidence. I suggest citing these, writing a paper about it, get it published in peer-reviewed journals, and THEN hope OTHERS consider it in an encyclopedia. —Preceding unsigned comment added by (talk) 19:01, 14 February 2011 (UTC)

Hrair from Watership Down[edit]

Does the term 'hrair' appear in Ed Yourdon's book? If somebody has access to a copy can they check. Is a term that appears in Watershipdown really of sufficient import to be mentioned in this kind of article? Derek farn 12:25, 11 September 2006 (UTC)

I've read Watership Down, and the book mentions the word and concept a significant number of times, and the reference could show acceptance or inspiration. I don't know if Watership Down or the paper came first, but one is likely based on the other and this would either show history or a popular view of the idea.joe conflo (talk) 00:10, 26 January 2008 (UTC)

How did Miller obtain 7+-2?[edit]

I've read Miller's paper multiple times, but I cannot find where he actually obtains a result of 7+-2. Some of the data he refers to falls far outside this range. I've even recall reading a slightly sarcastic analysis of Miller's paper saying 7+-2 should actually be 9+-4 from the data Miller presents. --Ronz 21:03, 7 February 2007 (UTC)

The reason you cannot find this result in his paper is that Miller never meant to stipulate a magical number. He has used this number only as a rhetorical device for the talk on which his 1956-paper is based (see Miller, G.A. (1989). George A. Miller. In G. Lindzey (Ed.), A history of psychology in autobiography (Vol. VIII, pp. 391-418). Stanford, CA: Stanford University Press). Therefore I would argue that “It supposedly [clarification needed] argues that the number of objects an average human can hold in working memory is 7 ± 2.” should read: “It is often misinterpreted as arguing that the number of objects…” I discuss this misinterpretation (including many references to peer-reviewed work) in more detail here: --HRLfeld (talk) 12:00, 20 April 2012 (UTC)

Difficulty remembering more than 7 unrelated nonsense data pieces[edit]

I'm just reading Tufte's book Beautiful Evidence now, and he sums up Miller's paper by saying "people had a hard time remembering more than about 7 unrelated pieces of really dull data all at once." Tufte emphasises that the experiment was about remembering nonsense data, and goes on to say that you can only reach the conclusion of "only 7 items belong on a list" if you don't read Miller's paper. This suggests that the current wiki-article's statement of "remarkable coincidences between the channel capacity of a number of human cognitive and perceptual tasks" -- to say that Miller is talking about cognitive capacity in general, and not just cognitive capacity when it comes to meaningless data -- is adding too many interpretations to this subject. Thanks. Chira 20:46, 30 March 2007 (UTC)

I don't understand how Tufte can make such conclusion. The experiments measured "absolute judgments." It's not nonsense, it's not meaningless data. Miller also discusses research into immediate memory. --Ronz 21:10, 30 March 2007 (UTC)
I think you misunderstand Tufte's point: it's not that Miller's data is nonsense or meaningless, it's that the "seven pieces of information" are nonsense data (random numbers, random sounds, random tastes, etc.). Tufte's point is that you can remember a lot more than 7 things if you contextualize the information; if it's just random chirps. If I quickly read off to you twenty interesting and previously unknown facts about people you knew in your life, obviously your brain wouldn't turn off after the eighth one. That's Tufte's point. Miller's paper tells us about the ability to remember a few, uncontextualized things at once; it isn't a general dictum on how much information the human brain can hold at any given time. -- (talk) 04:23, 25 January 2008 (UTC)
Miller's paper tells us more than that. Miller tells us (and Tufte tells us that Miller tells us;-) that, by adding additional, supporting dimensions of relationship (pitch, color, loudness, position in several axes, etc.) you can greatly multiply the "channel capacity" (immediate recall and identification). Whether the base article here handles Miller's thoughts well or not gets lost in a forest of technical terms, and questions about their general understanding. But it's certainly the case, as Tufte points out (and Miller has himself, widely) that Miller's work is not merely misapplied when used simplistically, its bulk and point is actually being ignored. Jackrepenning (talk) 23:36, 3 March 2008 (UTC)

Phone number length[edit]

Is this the study that is frequently cited for why US phone numbers were seven characters (and later, digits) long? --Mdwyer (talk) 21:00, 29 January 2008 (UTC)

There doesn't seem to be any documentation that AT&T ever did or used such a study. (talk) 23:40, 3 March 2008 (UTC)

No, he didn't![edit]

This article says (tail of the initial summary paragraph):

Miller hypothesized that these may all be due to some common but unknown underlying mechanism.

But Miller's article concludes:

What about the seven-point rating scale, the seven categories for absolute judgment, the seven objects in the span of attention, and the seven digits in the span of immediate memory? For the present I propose to withhold judgment. Perhaps there is something deep and profound behind all these sevens, something just calling out for us to discover it. But I suspect that it is only a pernicious, Pythagorean coincidence.

This entire article covers only the "urban mythos" arising from the first quarter of Miller's paper (and widely hung on the famous line which is its title). Should we write a real article on Miller's paper, or merely recategorize this one as covering the myth? Jackrepenning (talk) 00:16, 4 March 2008 (UTC)

Hrair limit[edit]

Is it just me, or is this utterly irrelevant to this article? Chris Cunningham (not at work) - talk 15:51, 22 March 2008 (UTC)

It is utterly irrelevant. But I have never been sufficiently motivated to delete it. Don't let me get in your way. Derek farn (talk) 19:42, 22 March 2008 (UTC)
I'm not sure it's so irrelevant. It's just another name for the same concept. I actually knew the term 'hrair limit' (which is more concise anyway) than 'magical number seven plus or minus two.' joe conflo (talk) 20:37, 23 March 2008 (UTC)
It isn't the same concept. 7+/-2 is an observed limitation strictly applicable to human short-term memory. The second appears to be a wider psychological observation. Maybe it needs its own article, but the hrair limit as described is not the same phenomenon. Anyway, this article technically concentrates on the paper, not on the phenomena itself. Chris Cunningham (not at work) - talk 16:48, 24 March 2008 (UTC)

Since the article no longer mentions the term, hrair should probably redirect to some other more applicable article, or risk confusing the hell out of everybody. — Gwalla | Talk 02:48, 10 November 2008 (UTC)

Remove Farrington link?[edit]

I agree that this citation is not the most relevant on the topic. For the misinterpretation issue, I would suggest to cite Miller himself: Miller, G.A. (1989). George A. Miller. In G. Lindzey (Ed.), A history of psychology in autobiography (Vol. VIII, pp. 391-418). Stanford, CA: Stanford University Press
As to the question of current estimates of working memory capacity limits around 4, I consider Cowan (2001) the most relevant paper. (This is already in the reference list.) Also Luck & Vogel (1997) is often cited (for example by Cowan, 2001). I think using the term „demonstrated“ for this point is fine, as a bunch of recent studies have shown working memory capacity limits around 4.
By the way, I do not think that the author really wants to say that „recent research has demonstrated that . . . the "law" [is] based on a misinterpretation of Miller's paper.“ How could research (other than historical) demonstrate that? Furthermore, there is not much left to demonstrate as Miller himself has already noted that he was misinterpreted (Miller, 1989).
--HRLfeld (talk) 12:33, 20 April 2012 (UTC)

Bit of nonsense about bits --> Unfortunately accurate representation of article[edit]

Hmmm... I noticed a bit of nonsense on display in this: "The information contained in the input can be determined by the number of binary decisions that need to be made to arrive at the selected stimulus, and the same holds for the response."

I thought, 'Heck, this can't be what he said. Obviously selecting between six, or seven, or eight different stimuli is NOT about making a series of "binary decisions."' But, unfortunately, upon review of the report I find that he pretty much did say this.

He also makes reference 2.5 bits, etcetera. I think, in attempting to explain things in laymen's terms, he engaged in very fuzzy and approximate descriptions. To say, for instance, that it takes 3 bits of data to represent 8 possible states (true) is IN NO WAY the same as saying that we need to make 3 binary decisions to identify a state (false). In fact, the rendering of data into binary is just a bit of silly number fiddling that obfuscates rather than illustrates.

There's a lot of other sloppiness in his presentation. For instance, what he said about "equally likely" alternatives is a bit of nonsense, which he again probably just inherited from an information-theory study where it was relevant and transported into an analogy where it is irrelevant. For example, how many bits of information is needed to represent whether a man is six feet tall? He says, if it's a 50-50 proposition, it takes just one bit to represent this information. Actually, it would take one bit to represent this information even if it were a 30-70 proposition. Or a 0-100 proposition.

So, anyway, no need to edit based on this info. I discovered that the article represents the original report. Surprisingly so.

Assessment comment[edit]

The comment(s) below were originally left at Talk:The Magical Number Seven, Plus or Minus Two/Comments, and are posted here for posterity. Following several discussions in past years, these subpages are now deprecated. The comments may be irrelevant or outdated; if so, please feel free to remove this section.

I have always been tought that empirical it is 3/4 chunks of data and not 7+-2 anymore, I am sure there is some prove for that somewhere.

Last edited at 09:11, 31 July 2007 (UTC). Substituted at 08:15, 30 April 2016 (UTC)