Talk:Feedback/Archive 1

Archive 1

on mention of "audio feedback"at top

postglock- lemme know if my last version works for you. thankz, Ished-out amounts of Vonn-ness 14:19, 10 September 2005 (UTC)

Proposal to refactor content

There's a long laundry list of important theories, phenomena and more built directly on top of feedback loops that aren't mentioned on this page but should, imo, be considered. (I think I could easily list a dozen, and maybe a hundred, but the specifics aren't my point and I really don't want to get into that game yet if we can avoid it.)

Conversely, much of what's on the page right now seems too detailed, even for me, someone who's been intellectually focused on feedback loops for over a decade.

I propose I/we refactor the page. The introduction should be as general as possible. A History section seems a natural. Maybe an Examples section. I see most of the details of particular feedback phenomena going on their own pages.

Comments?

love, raiph 03:44, 3 October 2006 (UTC)

Problems with the current definition of "Feedback"

> Feedback is the signal that is looped back to control a system within itself.

The mention of "signal" and "control" seem overly teleological. Here are some definitions of feedback loop from around the net; a definition of feedback ought to work well with these sorts of definitions:

• "A feedback loop is a system where outputs are fed back into the system as inputs, increasing or decreasing effects." (from this wikipedia's "feedback loop" article)
• "a linkage of two or more system components that forms a round-trip flow of information."
• "the output of some process or system is fed back into the system as a partial input, modifying the process."
• "a self-perpetuating mechanism of change and response to that change."
• "A series of actions which affect one another in a way that circles back to the original action."
• "In the climate system a “feedback loop” refers to a pattern of interacting processes where a change in one variable, through interaction with other variables in the system, either reinforces the original process (positive feedback) or suppresses the process (negative feedback)."
• "In a system where a transformation occurs, there are inputs and outputs. ... In every feedback loop, as the name suggests, information about the result of a transformation or an action is sent back to the input of the system in the form of input data."

What is your favorite, least constraining definition of feedback?

love, raiph 05:11, 10 October 2006 (UTC)

Merge Feedback loop into Feedback

• Support: It's the same content. Certainly feedback loops are everywhere, so this is in a way a disambiguation page.+mwtoews 21:40, 22 December 2006 (UTC)
• Support: The concepts are directly related. Zarggg 19:22, 1 January 2007 (UTC)
• Support: Same stuff --Pgreenfinch 19:44, 1 January 2007 (UTC)
• Lets do it: see above Sancho McCann 18:20, 3 January 2007 (UTC)
• Support; the other article has a pretty little picture, and links bordering upon the metaphorical, but it all belongs here. Jim.henderson 21:06, 24 January 2007 (UTC)
• Support: Presumably it's standard practice when making major changes like this to ensure the new page reads smoothly as a link from other Wikipedia pages. I'm thinking in particular of Cybernetics, which is basically the study of (higher order effects of) feedback loops. love, raiph 18:14, 7 February 2007 (UTC)
• Comment: It appears that the redirection has been put in place, so I removed the Merge discussion header. Nitsky416 05:47, 5 March 2007 (UTC)

Other Uses of the Word "Feedback"

There is a Rush album called Feedback that should be included in the disambiguation page. --The Fwanksta

The winning contestant from Who Wants to be a Superhero was called Feedback (Matthew Atherton). This should be included on the disambiguation page.

Feedback is also widely used to refer to suggestions given to people or companies in order to help improve behaviour, services, results or products. Feedback to students about homework results, to workers about team performance, to companies about product quality and so on. There is literature and research about effective feedback, impact of feedback and so on. It appears that Wikipedia does not have articles about this kind of feedback, other than 360-degree feedback - josei 19:06, 21 August 2007 (UTC)

Bogus examples

The example of vacuum advance control in older automotive ignition systems is bogus. Such systems did not employ any kind of feedback mechanism; i.e., they did not sense or react to any change they produced. —Preceding unsigned comment added by 24.22.7.221 (talk) 21:31, 17 January 2008 (UTC)

Incorrect sentence

The overview starts with the following sentence:

• Feedback is both a mechanism, process and signal that is looped back to control a system within itself

both is followed by three items instead of two. Hence, both should be removed. [Georg] —Preceding unsigned comment added by 206.113.192.12 (talk) 17:04, 25 January 2008 (UTC)

Ok, go ahead. -- Mdd (talk) 21:56, 25 January 2008 (UTC)

Content for first paragraph

I am used to finding concise explanations in the first paragraphs of a Wikipedia article, and would propose the following thumbnail sketch of "feedback" for the first para. I haven't edited an article so won't presume, but this example is technically accurate and memorable:

Feedback: an action changes a system and then the system changes the action.
Positive feedback increases the action: you step on the brake pedal in a moving car, and the car slows, pushing your foot down harder on the brake.
Negative feedback decreases the action: you step on the gas pedal and the car accelerates, pulling your foot off the gas pedal.

If this is appropriate perhaps someone will include it. 68.98.117.241 23:25, 27 January 2007 (UTC)

---

I agree with the above change, and suggest that the current definitions in Types of Feedback are in fact incorrect and misleading. An example of negative feedback is a thermostat heating a room that is too cold. This is not reducing output, but increasing output (from the heater) or increasing input (into the room).
I also feel that the "bipolar feedback" definition is confusing - both positive and negative feedback mechanisms can cause output to change in either direction, so both are bipolar. But feedback can't be positive and negative at the same time.
Finally, hunting is more correctly a case of over-control, where a negative feedback signal overshoots the mark. This is a type of oscillation, but doesn't strictly involve positive feedback at all.
I suggest that this article could benefit from better clarity of the terms positive and negative earlier on. I had started to edit, but realized that my background (electronics) is probably too narrow for a general article.
Trevithj (talk) 20:56, 18 August 2008 (UTC)

Bipolar feedback

This seems confusing. How can feedback be both positive and negative at the same time? These are exclusive, surely. If this means a switching function, where the feedback changes between positive and negative, then the definition should clarify this. If it means a balancing, then two feedback loops must be involved (one positive, one negative.)
I have searched several online sources, and they all seem to cite each other (including this one) but I can't find an original definition. I have noted several references to "bipolar feedback transistors", but suspect that means a feedback loop involving a bipolar transistor, rather than "bipolar feedback" involving a transistor.
If bipolar feedback is present in many systems, can someone provide an example?
Trevithj (talk) 02:43, 10 September 2008 (UTC)

Citations

Hello. This article has two references, which is good. I added a "citations missing" tag because most sections have no inline citations and thus there is no way for the reader to know to which of the two refs any particular piece of information is sourced. OK in advance from my point of view to remove or move or change this tag. Best wishes. -Susanlesch 18:43, 12 November 2007 (UTC)

Tahnks, I think there are no references at all in this article. So I replaced this tag. - Mdd 20:00, 12 November 2007 (UTC)

The concept of "feedback" is such a broad and high level concept that the primary purpose of this article is to provide an overview and to lead the reader to links that get deeper into more specific topics. Since wikipedia provides extensive in-depth details elsewhere, it would be a complete waste of time to replicate a huge list of references here. Please don't be simple minded trying to apply rules about citations, it is of benefit to no one. If there are examples where the information on this page disagrees with other more in-depth articles in Wikipedia then that should be fixed. 10:11, 5 October 2008 (UTC) —Preceding unsigned comment added by 203.206.220.108 (talk)

A possible article splitt should be cancelled

A splitting of the article has been proposed on 15 March 2007. Nobody made any suggestions about this untill I yesterday removed the splitting tag. Now today User:Intgr reversed this with the argument:

• I don't think it's outdated: the article still documents different loosely related meanings for "feedback" -- different concepts should have different articles

Now I think there are at least four good reasons not to split:

• A splitting proposal should be cancelled if nobody reacts. You can keep those ideas for ever.
• A splitt is not a good idea because:
  • there are no real different meanings of feedback in cybernetics, mechanical engineering and in economics.
  • They are all kind of control theory related.
  • Maybe only the super expert see all different things.
  • The article is about strong related meanings
• The fact that a band exist called Feedback (Matthew Atherton) shouldn't be a reason to make a disambiguation page
• I agree with josei. There is yet only a 360-degree feedback article, which can exist beside this article.

I think this article is only in need of an expert, who creates a more general introduction, which shows how all these aspects are related to one central concept. For example this article is comparable with the complexity article. This is also one paradigm, with different applications in the disciplines. - Mdd 18:53, 28 August 2007 (UTC)

The fact that nobody has done the split yet doesn't mean that it shouldn't be split at all. If the tag hasn't been discussed or contested since March then I'd think that most passers-by have been agreeing with the split, but haven't really bothered to do it themself.

My rationale: As you stated, the common meaning of "feedback" is its meaning in control theory — this which also embraces the positive/negative/bipolar feedback types. The overlap of how feedback applies to other fields (acoustics, optics, electronics, economics, nature or organizations) is strictly limited to the control theoretical meaning, so there is little reason to keep some of them together. In fact, almost universally, when you want to link to an article on "feedback" from another, you want to link to the term as it applies to a particular field.

The current sections on the fields in this article could surely be expanded, but right now they're stuck in a no-man's land: they do not form a part of a larger article, but they're not first-class articles either. You're saying that you are looking for an expert on this topic, but you will indeed need a "super expert" — rarely does a specialist grasp more than one of these fields.

So, as far as I can tell, this fits the pattern for disambiguation pages, and there are heaps of precedents like this.

Your mentioning of 360-degree feedback is inexhaustive: There also exist articles on audio feedback, optical feedback and low-key feedback. And the electronics meaning (which is currently a section in this article) even has its own category: Category:Electronic feedback.

My proposal: Either keep the control theory stuff here, or move to feedback (control theory) or some such. The rest should be split into feedback (electornics), feedback (biology) and feedback (economics) respectively; not quite sure what to do with the "organizations" section. -- intgr ^#%@! 22:12, 28 August 2007 (UTC)

Their is no need to spread so related knowledge over a number of small articles - Mdd 05:49, 29 August 2007 (UTC)

As I stated above, the only thing related between them is the control theory aspect. This concept will remain as a separate article, and will not be "spread" over the splitted articles. The fact that they currently exist as sections here actually limits their growth, since more specific content is irrelevant to feedback in control theory, even if it would be relevant in the particular field; just see the nature section: it briefly lists several phenomena without much further explanation, but is quite long already.

If there is no need to "spread knowledge", would you for instance consider merging audio feedback, optical feedback, and articles in Category:Electronic feedback here? Obviously not. -- intgr ^#%@! 08:26, 29 August 2007 (UTC)

Their are more arguments agains spliting the article:

1. The current article offers far to little information to justify the creation of separate articles
2. International there is also one article; See for example in the French, German and Dutch articles
3. An alternative can be an Feed back (disambiguation) page.

I also think that:

1. This article should first be expended and improved
2. Links should be made here to audio feedback, optical feedback and Electronic feedback

Now I think I look at the concept of feed back from a kind of fundamental "systems science point of view". In this perspective the control theory aspect is not whole separate aspects beside e.g. biological, economical, electronical, ergonomic and mechanical forms of feed back. The control theory aspects with their mathematical forms of representation are the fundamentals for all kind of specialized applications.

I do agree something has to be done with this article. In it's current form it's just a collection of seemingly unrelated facts. I'll see if I can improve this a little. - Mdd 20:14, 7 September 2007 (UTC)

Tag to split this article removed

I removed the tag to split the article because there was no response the last four weeks. - Mdd 19:59, 12 November 2007 (UTC)

Sorry, I just ran across this article now. This definitely needs to be split. As somebody mentioned above, the only thing linking these things together is a really broad concept of the term "Feedback". The methods by which feedback occurs within each of these fields - electronics, biology, communication - are really distinct, and feel haphazardly jammed together in one article like this. Put it this way: nobody is going to be looking for information about "feedback" at such a macroscopic level as to benefit from a one-size-fits-all article like this; they're probably going to be looking for it as it relates to a specific field. I can take this on as a personal project to split these off and build a worthwhile disambiguation page. Torc2 (talk) 21:57, 28 November 2007 (UTC)

Not haphazard, the article header explains it. Feedback is when the output of come process loops around back to become an input to the process. It simply is a broadly applied principle. —Preceding unsigned comment added by 203.206.220.108 (talk) 10:14, 5 October 2008 (UTC)

In electronic engineering

The subsection "In electronic engineering" re-explains the concepts of negative feedback and positive feedback, which are not particularly specific to electronic engineering. Anyway, the section that contains this subsection is "Applications" so this is not the right place for explaining a general principle. 203.206.220.108 (talk) 01:28, 8 November 2008 (UTC)

Feedback equals No Feedback

• 3 - the feedback signal does not change when the input signal changes, in which case there is no feedback. This is known as a Linear system.

If no feedback is an example of feedback then sitting on your bum is an example of flying. In addition linear systems can have feedback, and non-linear systems do not require feedback. Thus the above should be deleted. 203.206.220.108 (talk) 12:31, 28 December 2009 (UTC)

Bad example, When everytime I've been on an airplane, I was sitting on my bum. —Preceding unsigned comment added by RaptorHunter (talk • contribs) 02:37, 21 November 2010 (UTC)

This article suffers from vagueness

This article has the same naming problem that correlation article had years ago. People at first mislabeled correlation as positive and negative. That is nebulous. It was correctly changed to direct and inverse correlation.

This article suffers from the same issue. It should NOT be positive and negative feedback. Instead the proper more accurate terms are converging and expanding feedback. The article needs renamed as well.

173.188.7.24 (talk) 21:19, 4 March 2012 (UTC)

Please provide sources for the terms "expanding feedback" and "converging feedback" and why they are more applicable than the traditional terms, positive and negative feedback. Jojalozzo 22:10, 4 March 2012 (UTC)

May I suggest that the terms "reinforcing loop" and "balancing loop" are used more often to avoid the ambiguity of the terms "positive" and "negative" (and even "feedback"). See for example the work of John Sterman and Peter Senge, and Systems dynamics or Causal loop diagram. Trevithj (talk) 03:18, 7 March 2012 (UTC)

They could be mentioned, but probably not "used more", since they aren't the terms most commonly used in discussion of feedback. Dicklyon (talk) 07:11, 7 March 2012 (UTC)

That may be true, but I submit that "reinforcing loop" is a less ambiguous term than "positive feedback". Both "positive" and "feedback" have (by my count) three distinct meanings, so the common term could refer to "desirable speaker-howl", "beneficial appraisal" or "increasing criticism", to name a few. Trevithj (talk) 01:51, 8 March 2012 (UTC)

Types of feedback

I suggest that the primary purpose of the "Types of feedback" section is to distinguish negative feedback from positive feedback and outline other commonly used terms for types of feedback. This section should not get bogged down into fine details. The material that discusses the history of negative feedback should be moved into the Negative Feedback article. 203.206.220.108 (talk) 00:44, 8 November 2008 (UTC)

However, the origin of the term "feedback" may be relevant, if that is what the Black and Rosenblueth citations are trying to do. Maybe some clarity is needed here, since the original use of "feedback" seems to have described what we now call "audio feedback", which is a more specific term.

On the other hand, there is a lot of emphasis put on the electronics POV. For example, the assumption that feedback goes from output to input (vs Ramaprada's generalization that can apply to any system parameter). Useful for the origin of the term, but no longer generally true. Also, the mechanism of applying feedback is assumed to be by cancelling or reinforcing some of the input signal - again very specific to electronic signal amplifiers.

I support the above idea, and further suggest several points in this section be made more generic. To that end, figure 1 should be moved down to the Electronics section. -- Trevithj (talk) 23:51, 20 March 2012 (UTC)

Clarity on Usage of Positive/Negative

As a suggestion, I would like to add a lead to the Types of Feedback section ... something like this:

Feedback is commonly divided into two types - termed positive and negative. How these terms are defined can vary depending on the context, and on the reference value. It is possible for a specific case of feedback to be defined either as positive or negative, depending on how it is measured. [Levine et al., 1992]

Positive and negative feedback can refer to:

1. the gap between reference and actual values of a parameter, based on whether the gap is widening (positive) or narrowing (negative).[A Ramaprasad 1983]
2. the desirability of the means of correction, based on whether it is pleasant (positive) or unpleasant (negative). [BF Skinner 1954]

-- Trevithj (talk) 20:48, 22 March 2012 (UTC)

Initial modifications made. I intend to remove the paragraphs in this section that deal with the electronic amplifier example, and the positive/negative feedback paragraphs too. They seem very specific to electronic circuits, and are largely a repeat of what is already stated elsewhere in the article.

--Trevithj (talk) 05:57, 29 March 2012 (UTC)

Tidying the Overview

I see a need to make this section:

• more generic. The current overview emphasises the cybernetic/electronic sense, jumps too quickly into definitions of positive/negative, and assumes a mechanism of control that is largely specific to electronics (output used to influence input). Feedback doesn't always involve measuring 'output', and not all mechanisms influence 'input'. (Ramaprasad, 1983)
• more about the origin of the term. The paragraphs on feedforward seem to be trying to do this, but don't define feedforward clearly (with citations). Also other authors use 'feedforward' in a different way to this (e.g., Hendry 1988), which is confusing.

I propose: describing the history of the term "feed-back", and show its adoption from verb phrase to noun; moving the examples to their relevant sections in Applications, where they can be explained in greater detail; removing the description of positive/negative types, since these are already covered later in the article - and have their own pages anyway. --Trevithj (talk) 22:09, 29 March 2012 (UTC)

Remove redundant synonyms

Any objections if I remove the Audio feedback and Performance appraisal sections in the lead? They aren't really in the right context here, and are already covered in the disambiguation page and the "See also" list. --Trevithj (talk) 00:43, 18 April 2012 (UTC)

Generic diagram for lead

Proposal is to return the "ideal feedback model" image to the electronics section, and put a more generic image in the lead. The existing image IMO better serves the article if placed in the Electronics section, for reasons stated here.

I agree with User:Teapeat that the lead image needs to summarize the overall topic in an abstract way. The image I propose follows Senge's definition of feedback as "any reciprocal flow of influence ... every influence is both cause and effect", or Astrom & Murray's "situation in which two (or more) dynamical systems are connected together..." and would consist of two abstract entities connected to each other by arrows representing a loop of influence or causality. This is equally applicable to either type of feedback, and applies broadly to any example.

Other options are welcome. --Trevithj (talk) 22:20, 19 April 2012 (UTC)

Note that source of the diagram (as "ideal feedback model") appears to be Wai-Kai Chen, Active Network Analysis (1991, pp186-187), who likely based it on Hendrik Bode's Network Analysis and Feedback Amplifier Design. Can anyone sight and cite Bode's book? -- Trevithj (talk) 21:44, 7 May 2012 (UTC)

Okay, after a month and no comments, I will move the "ideal feedback model" diagram back down to the Electronics section, with cited reference to (and diagram description from) W.Chen. -- Trevithj (talk) 23:13, 17 May 2012 (UTC)

Sorry for the shortage of feedback to your suggestions. Sometimes it's just easier to react to edits, rather than to ideas. I'm not so sure; the system diagram is not particular to electronics, though it is most typically studied in EE courses. What diagram are suggesting to put up front instead? I think something with an input and and output makes more sense than the abstract of two things feeding back on each other.

There is an old conflict of definitions here - check out the quote in the Feedback#Overview section. My background is in electronics, and I took for granted that the output is used to modify the input. But that actually is very specific to electronics, where two signals can cancel or enhance each other. Get away from electronic signals, and the model doesn't hold. Also, I couldn't find a single cited source for the previous diagram that didn't involve electronics. -- Trevithj (talk) 23:57, 17 May 2012 (UTC)

Maybe not cited, but this book has block diagrams with outputs feeding back to summing nodes at inputs, and discusses it as such. Dicklyon (talk) 01:34, 18 May 2012 (UTC)

I don't know the book, but if it is about the origins - that would very likely include Black's negative feedback amplifier, as discussed in the Feedback#Overview.-- Trevithj (talk) 02:04, 18 May 2012 (UTC)

As for dividing feedback into two types, positive and negative, that's too simple in general. Real feedback systems tend to be a mix, depending on frequency and such. Dicklyon (talk) 23:30, 17 May 2012 (UTC)

Frequency? Would I be right in assuming you are thinking of electronic signals here? -- Trevithj (talk) 23:57, 17 May 2012 (UTC)

Frequency is important in all linear systems, and most nonlinear ones, too. My background is also EE and electronics, but since I work on modeling hearing, the system I'm working with is not electronic, but full of feedback of various sorts. Dicklyon (talk) 00:52, 18 May 2012 (UTC)

Oh. Okay, sorry for presuming. Re the two types - they are commonly used in the literature. I agree that real systems are a lot more involved (quote by WR Ashby "The fact is that the concept of "feedback", so simple and natural in certain elementary cases, becomes artificial and of little use when the interconnexions between the parts become more complex.") but I figured we should follow the peer-reviewed majority for this article. -- Trevithj (talk) 02:04, 18 May 2012 (UTC)

Cited Definition

If no objection I would like to insert the below cited definition into the preamble, just above "feedback is also a synonym for...":

Feedback is information about the gap between the actual level and the reference level of a system parameter which is used to alter the gap in some way.

[Arkalgud Ramaprasad, "On The Definition of Feedback", 1983.

I hesitate only because the definition comes from the Behavioral Sciences, and may not be broadly applicable. Although it looks pretty good to me. --Trevithj (talk) 22:04, 15 March 2012 (UTC)

Above changes made, and also modified the synonym "feedback signal", since this definition makes that (old) distinction unnecessary/redundant. --Trevithj (talk) 05:51, 29 March 2012 (UTC)

I would say that's a very odd and specific definition, not very applicable to the general case in engineered systems, where there is often no "reference level" or "gap"; a simple definition from a dictionary such as this one might be better. Dicklyon (talk) 23:21, 17 May 2012 (UTC)

Thanks for the definitions. I had considered something similar. Notice that the two Science and one Behavior defs refer to "information". In general, this information is often taken to imply a "discrepancy".

I'm a little surprised to hear that there is often no reference level. How does an engineer know when the system is working correctly? Can you give an example? -- Trevithj (talk) 02:18, 18 May 2012 (UTC)

Email and social science

Is email administration really a social science? Maybe in the 90's it was, but we must have passed that stage. — Preceding unsigned comment added by 14.200.55.92 (talk) 08:08, 30 November 2013 (UTC)

Types of feedback

The "Types of Feedback" section was poorly written, and jumped immediately to a rather unclear discussion on the social-science meaning of the words "positive feedback" and "negative feedback". I've tried to rewrite it to make it a bit clearer. Geoffrey.landis (talk) 16:12, 27 June 2014 (UTC)

I don't question your intentions. However, the terms "positive" and "negative" are ambiguous terms, and the previous version reflected this. It didn't say anything about social science - in fact the quoted definitions come from behavioral and management science. So now one context is assumed, and the other is rather narrowly labeled almost as an afterthought, whereas before they were given a more equal weighting. New version isn't really clearer. Trevithj (talk) 09:20, 28 June 2014 (UTC)

However, your approach in moving the discussion on conflicting definitions out of the discussion of types does make sense. I have generalized the sub-heading (it may well be a social-science phenom, but citations are required.) Sorry if my edit-reversal was too precipitous. Trevithj (talk) 21:46, 2 July 2014 (UTC)

the "types of feedback" section is still rather poorly written. "the altering of the gap between reference and actual values of a parameter, based on whether the gap is widening (positive) or narrowing (negative)" may be technically accurate, but it is nearly incomprehensible to anybody who doesn't already know what feedback is. Geoffrey.landis (talk) 17:20, 10 August 2014 (UTC)

Definition appears to be over narrow, propose new definition

The current definition used in the article to scope it is:

Feedback is a process in which information about the past or the present influences the same phenomenon in the present or future.

The issue I have with this is that it doesn't map well onto many feedback loops. For example a ball-cock/float valve does not communicate information in any normal sense, it physically closes the valve. Another example is climate, there are feedback loops on the rainfall and temperature that go via bacteria seeding clouds. These are not information in any normal sense. Information is usually considered as a trigger to action, whereas these are mechanical feedback loops that physically force negative feedback.

Additionally there is the question of what the 'same' phenomenon means; it could mean practically anything.

I propose to change it to:

Feedback is any process in which the state of a system influences the future state of that system via a closed causal loop.

This is slightly more abstract, but the concept of the 'state of the system' is fully general, as is the term 'influence' and the notion of the loop being closed and causal. These seem to be wider than the current definition and the minimum necessary ingredients for feedback to be said to occur.

I welcome comments.GliderMaven (talk) 16:10, 10 August 2014 (UTC)

I think that making up abstract definitions without showing support for them in sources is not likely to get us anywhere. And I think your definition is too general, in that the state of a system almost always has a causal influence on future state and output, whether there is something identifiable as feedback or not. The previous definition has the same problem, assuming you interpret information broadly enough for it to mean anything. Dicklyon (talk) 16:53, 10 August 2014 (UTC)

In a ball-cock valve, the "ball" is a device which measures the amount of water in the tank. This information is fed (via a mechanical connection) to the valve, the device which controls the rate at which water flows to the tank. In modern times, the idea that information can be fed via a mechanical connection seems old-fashioned, but it's still feedback. Geoffrey.landis (talk) 17:17, 10 August 2014 (UTC)

Unless you are getting into quantum mechanics, which I don't think we are, information is a human thing, whereas the actual feedback is a mechanical or electrical or chemical causal connection.GliderMaven (talk) 18:52, 10 August 2014 (UTC)

Dicklyon has pointed to what is the point of WP: to report the contents of reliable sources and not WP editors' personal and unsourced views. If a definition is to be discussed, it should be related to sources. It doesn't have to copy them verbatim, of course, but it should be a clear presentation of the content of sources. Brews ohare (talk) 18:37, 10 August 2014 (UTC)

On the contrary, this is very often misunderstood, and you have done so. The first few sentences of the lead is the part where we state what the editors are considering to be the scope of the article for the readers; what is, and is not in scope is firmly an editorial decision. It cannot be found in any reference at all.GliderMaven (talk) 18:48, 10 August 2014 (UTC)

Well, I doubt you believe the intro is open to wild imaginings. Such invention is constrained where definitions are involved by citing supporting sources, a common practice on WP that does not interfere in any way with the purpose of an intro. However, if you object, then provide the sources on this talk page so editors can assess that your proposed definition is indeed consistent with a balanced view of published reliable opinion, and is not a poorly worded approximation. If editors find your sources provide good support, they can decide whether to cite them in the lede or not. Brews ohare (talk) 19:24, 10 August 2014 (UTC)

That's basically what I'm doing here, in marked contrast to what you seem to be up to- which is forcing some randomly chosen definitions references on Wikipedia. The point is that the editors, particularly with the wider scoped, more general articles such as this one, end up having to synthesise an umbrella definition that encompasses all of the other definitions (or most of them, some definitions when you look at them hard enough are simply incorrect; dealing with incorrect reference information is something else that Wikipedia editors have to deal with.)GliderMaven (talk) 14:17, 12 August 2014 (UTC)

Ashby's discussion of the limited utility of 'feedback' as a concept

Before becoming embroiled in extreme positions on what is or is not feedback, it is worthwhile to consider Ashby's discussion of the limitations of this concept.

In Ashby's extensive work, on p. 58, article 4/10 he compares the case of two parts of a system, P and R and contrasts the case where P affects R but not the contrary, denoted P → R, with the case where each affects the other, denoted P $\rightleftarrows$ R. In this last case he says "when this circularity of action exists between the parts of a dynamic system, feedback may be said to be present." He continues:

"Other definitions, however, are possible, and there has been some dispute as to the best."

Among these, he considers those

"who want to use the word to refer, when some forward effect from P to R can be taken for granted, to the deliberate conduction of some effect back from R to P by some connection that is physically or materially evident."

He goes on to say that for the mathematician the requirement of a physical connection renders the theory "chaotic and riddled with irrelevancies". Ashby continues:

"The exact definition of 'feedback' is nowhere important...the concept of 'feedback'...becomes artificial and of little use when the interconnections between the parts become more complex. When there are only two parts...the properties of the feedback give important and useful information about the properties of the whole. But when the parts rise [in number] to even as few as four, if every one affects the other three, then twenty circuits can be traced through them, and ...[they]... do not give complete information about the system. Such complex systems cannot be treated as an interlaced set of more or less independent feedback circuits, but only as a whole...For understanding the general principles of dynamic systems, therefore, the concept of feedback is inadequate in itself...Complex systems ...have complex behaviors, and these behaviors can be goal-seeking in complex patterns."

The implications here are that feedback has a number of possible definitions, and whatever definition is chosen it will prove immaterial to the understanding of complex systems. This idea becomes paramount in philosophical considerations like enactivism, emergence, self-organization, and situated cognition.

With that in mind, a view of feedback in this article that takes a somewhat skeptical and distanced view of the matter seems appropriate, presenting a variety of views and their limitations. Brews ohare (talk) 22:08, 10 August 2014 (UTC)

This work is exceptionally dated. He's talking about 4 feedback loops.

The computer I have in front of me has of order a thousand million feedback loops inside it. A lot of that is RAM, but the microprocessor also contains large numbers of them; and the basic operation of the microprocessor actually is feedback. Computation is feedback, the program counters and registers are fed back internally in complex patterns over buses and so forth.

Far from being some little irrelevancy, the concept of feedback is completely central to computing.

The idea that coupled feedback loops can produce complex behaviours and so feedback is unimportant is exactly backward to the reality; coupled feedback loops are used incredibly, amazingly extensively for that very reason.

If feedback is so unimportant, why is it so important to my computer?GliderMaven (talk) 03:39, 11 August 2014 (UTC)

There are two problems here. The first is that you misunderstand Ashby, who is simply expressing the view that understanding feedback in simple cases is not helpful in complex cases where it is difficult to trace the interdependencies that may defy subdivision into parts with separable interactions. (The very simple case of feedback circuits in amplifiers is complicated in this way. See. for example, Chen and the article on return ratio, a method used because feedback is hard to quantify even in relatively simple circuitry.) Even where few parts are involved he has pointed out that debate exists. The more important problem is that you are not objecting to Ashby based upon sources with a less 'dated' perspective, but using your personal interpretation of your own experiences. Brews ohare (talk) 13:07, 11 August 2014 (UTC)

Things like Chaos theory postdate Ashby; and are formed from using feedback in non linear systems.GliderMaven (talk) 15:11, 12 August 2014 (UTC)

There's an implicit idea that all forms of feedback are positive or negative; this is clearly false. And it's not difficult to see that a point in a state space of system with 3 or more dimensions can be both positive and negative at the same time when perturbed in different directions, although that would normally be considered positive feedback behaviorally.

Another thing that's missing from the article is catastrophe theory; which leads to things like Schmitt triggers which also use feedback, but are not simple positive/negative.GliderMaven (talk) 15:13, 12 August 2014 (UTC)

Ashby advocates the "circular causality" idea - which I believe is the thinking behind GliderMaven's suggestion. While historically the term "feedback" has referred to "output-to-input" (or out→in for short) the causal loop idea is more generic - which is what we want here.

I can see that the information term could be confusing. If we are using a specific term in a generic sense, we could either make that clearer, or simply drop the term. I would argue that the same can be said for the term "process", which strongly implies out→in.

Regards causal loops: it seems we have two basic models: a system that influences itself, or two parts of a system that influence each other. (S→S or P $\rightleftarrows$ R) I suggest they are the same idea, with differing levels of detail. Which is the more convenient for purposes of explanation? Trevithj (talk) 19:58, 11 August 2014 (UTC)

Sorry Trevithj, I'm having trouble understanding your remarks which seem to assume some context I don't have. If you can understand GliderMaven, maybe you can also understand that if he used sources, his explanatory powers and credibility would be enhanced?. As for S⟺S vs. P $\rightleftarrows$ R, what is it you want to explain?

If you are talking about "information" as in "Feedback is the return of information about a system", this is from Morris, and I don't know what is confusing about it. Maybe you can elaborate? Brews ohare (talk) 21:22, 11 August 2014 (UTC)

Yes, that's the kind of thing. That's a more general definition, and this needs to be a general article that points off to the more specific places.GliderMaven (talk) 15:11, 12 August 2014 (UTC)

Ooops

Sorry about my broken edit, and thanks for fixing it, GM. I must have been looking at an old version.

On the Ramaprasad ref and definition, I don't know what to think. Does someone have a copy I can read? As an attempt at a "general" definition it seems way narrow, so it's not clear why we are including it there. Dicklyon (talk) 16:34, 26 September 2014 (UTC)

I have a hard copy, but am unable to find online free version. Wish it were more available - his references are very good, IMO. Can discuss, but am travelling at present. Trevithj (talk) 19:49, 26 September 2014 (UTC)

I bought a copy. I see it's all about the use of the term in "management theory", so I will clarify that in the article. Dicklyon (talk) 00:01, 27 September 2014 (UTC)

Here's the abstract:

The paper discusses the lack of a commonly accepted definition of the concept of feedback in management theory, dealing with communications networks and decision processes in living systems at the organization level. It proposes a general definition. It deals with living systems at the organization level, including the total system and all subsystems. Implications of the proposed definition to current conceptualizations of feedback processes in management are explored.

KEY WORDS: organization, decision making, feedback, management theory.

Thanks Dick for the summary. Consider this quote:

Thus as long as the definition of the concept of feedback used by management theorists is at variance with the definitions used in cybernetics, general systems theory, control theory, etc., knowledge from these disciplines cannot be translated into the context of management. ... On the other hand, there is no universally accepted definition of feedback outside management theory either.

The second point is the difficulty we are having. While focused on management theory, the usefulness of this paper is the author's broad discussion of definitions across many disciplines, and his rationale for choosing the definition he does. The exact definition may not be agreeable for this article, but I believe it to be a very useful guide for discussion. Trevithj (talk) 07:14, 27 September 2014 (UTC)

It also is the first result in Google scholar when I search on "definition of feedback", for what that's worth. Trevithj (talk) 09:04, 27 September 2014 (UTC)

I think if we make the (to some extent artificial if you believe in physicalism) distinction between feedback and reinforcement, then the problem goes away. If you subscribe to physicalism, there's still no strong necessary connection between the bulk of mechanistic feedback theory and reinforcement, due to the much higher complexity of animals, but in that case you could add reinforcement as a subarticle.GliderMaven (talk) 14:34, 27 September 2014 (UTC)

In terms of definitions, to the best of my knowledge, this diagram is basically the definition of feedback:

The general feedback loop where all outputs of a process are available as causal inputs to that process

where the inputs and outputs can be many, since any feedback system can be mapped to it, if P is allowed to be any general function of its inputs; any feedback system has an equivalent system of this form.

I saw the diagram in a book once, but I don't have a reference, I already added it to the article because it identifies the input output and system boundary for the existing text.GliderMaven (talk) 14:34, 27 September 2014 (UTC)

Negative feedback definition

At present, the lede states:

"The term "feedback" in this sense is also used as an abbreviation for:

• Negative feedback – when the fed-back information acts to control or regulate a system by opposing changes in the output or measurement.

This definition of negative feedback suggests that it is an event that occurs when the fed-back information acts. The idea that negative feedback refers to an 'event' seems awkward to me because it emphasizes a transient behavior, rather than the consequence of a particular arrangement or set-up. I am also disturbed by the notion of information as an agent that can 'act'.

Some sources refer to negative feedback as a process.1 2 3 or as a form of control,4, or as a particular circumstance.5.

That suggests this change:

"The term "feedback" in this sense is also used as an abbreviation for:

• Negative feedback – a process in which the fed-back information acts to control or regulate a system by opposing changes in the output or measurement.¹

Sources

¹Chris Impey. How It Ends: From You to the Universe. In negative feedback, a signal fed back into a system is reduced or canceled so that change is minimized

Of course, a different reference could be chosen. Brews ohare (talk) 17:25, 27 September 2014 (UTC)

There being no discussion here, I implemented this change with two additional references. Brews ohare (talk) 15:29, 29 September 2014 (UTC)

Deletion of large sections on feedback in education, management etc.

This edit removed extensive sections dealing with educational and other aspects of feedback without providing any feedback here on the Talk page. The in-line editorial comment was made regarding the educational material that article is not about individual uses of feedback, it is about feedback loops; the entire loop- and in quantitative science, and about the social science material removed the most egregious examples and unreferenced material.

No doubt, some changes were warranted. However, some discussion also is warranted. In particular, the idea that this article should be restricted to "quantitative science" is rather debatable, particularly because cybernetics arose in this context.

There appears to be a prejudice here that 'feedback' in other areas is used only in the everyday sense of rejection and unfavorable comment. However, that is not the case. Material where feedback is used to imply a detailed critique aimed at modifying a process or organization seems to me an appropriate topic for this article. While overlap will occur with the subjects of Reinforcement or "operant conditioning", and Reinforcement learning, the more technical aspects of feedback are mentioned in these fields and this article could usefully include some subsections that briefly make these connections. Brews ohare (talk) 17:56, 27 September 2014 (UTC)

Whether they're formally types of feedback or not, they're still firmly and clearly forms of reinforcement; if you want to cover them, you should cover them in the most relevant article, which this article isn't.GliderMaven (talk) 16:02, 28 September 2014 (UTC)

Reinforcement is an aspect of the use of feedback, but it is only part of it. And even if were everything, which it is not, this article should mention it as some readers will visit this page with applications to education and management in mind, and the connection with cybernetics, and wonder why it is not here. Brews ohare (talk) 00:32, 30 September 2014 (UTC)

Feedback mechanism

Brews, why the hangup on "feedback mechanism"? Isn't an X mechanism just a mechanism involving X? What needs to be said about that? And where did you get that crazy definition you had added? Dicklyon (talk) 00:42, 30 September 2014 (UTC)

Dick, your view that this is a 'hang up' is your own opinion favoring 'feedback' as a noun and denying its use as an adjective to identify a particular type of mechanism, afeedback mechanism. It appears some authors like the latter.

I'm not sure what 'crazy' definition you mean, and just why is it 'crazy', pray tell?

Dick, try addressing sources in place of me. Brews ohare (talk) 02:01, 30 September 2014 (UTC)

Feedback mechanism – the action or means used to act upon feedback information, to implement a response

is what you "restored". I don't recall seeing it before; maybe someone else made it up if not you. But it's lame.

and then you came back with "feedback mechanism" as a synonym for "feedback", and piled on a bunch of refs that use that term. Nobody is denying that such term is in use, with feedback as an adjective. That's why I wrote "Isn't an X mechanism just a mechanism involving X?" Well, isn't it? What is your point here? What is it about the term "feedback mechanism" that so makes you want to include it in the lead? Dicklyon (talk) 02:39, 30 September 2014 (UTC)

It's used by many authors - cited several because I wanted it clear when you reverted that you were censoring. Brews ohare (talk) 03:01, 30 September 2014 (UTC)

Censoring, right. It's in my book of prohibited concepts. No mechanisms. Dicklyon (talk) 04:05, 30 September 2014 (UTC)

Yes, feedback mechanism is common enough in sources. But what do you want to say about it that's worth a mention in the lead? So far I don't see anything. And I have no issue with sources, which is why I address you, Brews; tell us why you think this is worth a mention in the lead. Dicklyon (talk) 04:13, 30 September 2014 (UTC)

Older comments

is external linking allowed?, in any case much good info in http://prof.usb.ve/williamc/Mejoramiento/historia.html should be made available here

in electronics the para are disordered, positive feedback (suject of great Armstrong deForset patent dispute)predates Black's negative feedback amplifier and should be refered to first, PID is a subset of negative feedback control techniques :jcox

--- I don't believe that the explanation of audio feedback is correct. It identifies the frequency of feedback noise as related to the transit time through the system (Implying it's a matter of repeating the waveform, the effect you might get with digital echo or flange). But I'm 95% sure that's completely irrelevant in a system like this, the waveform is completely obliterated by all the transitions made, all you're left with is frequencies that approximate those you started with.

OK, a quick test: I was able to get audible feedback with a 1/2 second delay in the loop, which should give a frequency of 2 hz, 5 octaves below audible.

So I'm rather sure the frequency of feedback is not determined by transit time, but by the frequency response of the whole system. All real world systems contain some capacitance and inductance, meaning they all act as very wide, low, bandpass filters. But when you amplify and feedback the filtered sound into the filtered sound, you've naturally created an oscillator (the simplest forms of analog oscillators work on exactly this principle).

So unless anyone objects I'm going to change the explanation to reflect this. User:Wji

---

Although feedback was perhaps discovered during the course of the development of electronics, it exists in many natural processes and the word (although not present in earlier dictionaries, for example the original OED) represents a concept in general use outside engineering. Therefore the article should begin with the general concept but include a substantial section on control theory. I was first introduced to feedback during training on servo mechanisms many years ago and am a bit rusty, but I clearly remember it took both positive and negative feedback to make them work. What we call feedback like with a microphone is perhaps not feedback in the sense we are using it here. User:Fredbauder

I agree that there is more than one sense of the word feedback. The Penguin Dictionary of Electronics, for example, has separate entries for feedback in the general sense and feedback control loop in the specific context of control systems. The Chambers Science and Technology Dictionary has headings for feedback (acoustic), feedback (telecommunications) and feedback control loop, so perhaps there are really three different concepts. If we wish to separate the general from the particular, I would suggest the following structure for the Wikipedia (with separate articles for hunting and feedback):

• Hunting (known to mechanical engineers as early as 1880, according to OED2)
• Feedback
  • as originally defined, i.e. in electronic amplifiers (see OED2 citation from 1920)
    • acoustic feedback in public address amplifiers (OED2 has a 1921 citation of howl) - perhaps this could be a separate article
  • feedback theory applied to control loops (probably first applied in World War II)
    • control loops in nature, i.e. natural systems that can be analysed with control loop theory (existing since prehistory but only understood by humans in the light of control theory)

Do you think this is reasonable? Unfortunately it contradicts your idea of putting the natural phenomena first. My scheme lists things in the order of discovery by man, whereas yours is nature-based. I cannot say which is better.

As for the microphone example, this is an illustration of feedback in the electronic (or electroacoustic) sense, not in the control loop sense. The feedback is always positive, because sound from the loudspeakers is always added to, never subtracted from, the sound entering the microphone. What changes is the loop gain, which varies as the user moves the microphone relative to the speakers. As soon as the loop gain exceeds unity, the system howls. This effect is independent of frequency. I believe that a system with negative feedback can only become unstable if the signal frequency exceeds the bandwidth of the feedback loop.

In control theory you can have multiple feedback paths (often three: proportional, integral and derivative) and it is meaningless to talk about overall positive or negative feedback, because the three paths may have different characteristics. --Heron

What I suggest is a simple introduction which includes the general case, followed by as much specialized technical information as you feel appropriate. I agree the notion of postive and negative is applicable only to 2 dimensional systems, but I think it should remain in the introduction. I don't think organization around order of discovery is good; I prefer organization to start with a brief general introduction using simple language and concepts followed by (in this case) an engineering section, followed by the ways the concept has been adapted to natural systems. I guess runaway positive feedback is what we are talking about with apmplifiers. Fred Bauder

I divided the article into sections, without changing the words much. I may go further and remove the acoustic example to a separate article, but I'll leave 'hunting' where it is. --Heron

---

I have been bold and removed the acoustic feedback section totally. I have linked the existing article at Audio feedback to the disambiguation page. The main differences in the removed section (reproduced below) lacking in the original page are the points about the band-pass filter and the electronic oscillator.

I believe the band-pass filter point is incorrect anyway. True, all systems have some kind low and high pass filter in them, but this is generally well outside the human range of hearing, and in any case, makes no difference to the presence of feedback. The deleted section was also vague as to whether this was affecting the presence or nature of the feedback, but if it were just the latter, there are plenty of other factors involved (at the very least, the inherant EQ of the system would be much more important).

Also, the electronic oscillator point seems slightly amibiguous too, although I could understand if people wanted this link in the main article. It might be worth pointing out that the simplicity is certainly not in the waveform, but in the construction. --postglock 07:49, 6 Jun 2005 (UTC)

Feedback in audio systems

Main article: Audio feedback.

A well-known example of runaway positive feedback in electronic systems is called "howl" or "howl-round". This occurs in public address systems when sound from the loudspeakers reaches the microphone, is amplified by the system, and is then fed back into the system at even higher volume. All electrical systems contain capacitance and inductance and so act as band-pass filters responding better to certain frequencies. In a single loop through the system, this effect is negligible, but it becomes severe when the signal passes through the system repeatedly. This effect is the basis of the simplest kinds of analogue electronic oscillator.