Skip to main content

New research on understanding metaphors

September 29, 2008 by coglanglab

coglanglab's picture

Metaphors present a problem for anybody trying to explain language, or anybody trying to teach a computer to understand language. It is clear that nobody is supposed to take the statement, "Sarah Palin is a barracuda" literally.

However, we can imagine that such phrases are memorized like any other idiom or, for that matter, any word. Granted, we aren't sure how word-learning works, but at least metaphor doesn't present any new problems.

Clever Speech

At least, not as long as it's a well-known metaphor. The problem is that the most entertaining and inventive language often involves novel metaphors.

So suppose someone says "Sarah Palin is the new Harriet Miers." It's pretty clear what this means, but it seems to require some very complicated processing. Sarah Palin and Harriet Miers have many things in common. They are white. They are female. They are Republican. They are American. They were born in the 20th Century. What are the common characteristics that matter?

This is especially difficult, since in a typical metaphor, the common characteristics are often abstract and only metaphorically common.

Alzheimer's and Metaphor

Some clever new research just published in Brain and Language looked at comprehension of novel metaphors in Alzheimer's Disease patients.

It is already known that AD patients do reasonably well on comprehending well-known metaphors. But what about new metaphors?

Before I get to the data, a note about why anybody would bother troubling AD patients with novel metaphors: neurological patients can often help discriminate between theories that are otherwise difficult to distinguish. In this case, one theory is that something called executive function is important in interpreting new metaphors.

Executive function is hard to explain and much about it is poorly understood, but what is important here is that AD patients are impaired in terms of executive function. So they provide a natural test case for the theory that executive function is necessary to understand novel metaphors.

The results

In this study, AD patients were as good as controls at understanding popular metaphors. While control participants were also very good at novel metaphors, AD patients had a marked difficulty. This may suggest that executive function is important in understanding novel metaphors and gives some credence to theories based around that notion.

This still leaves us a long way from understanding how humans so easily draw abstract connections between largely unrelated objects to produce and understand metaphorical language. But it's another step in that direction.

-----
M AMANZIO, G GEMINIANI, D LEOTTA, S CAPPA (2008). Metaphor comprehension in Alzheimer’s disease: Novelty matters Brain and Language, 107 (1), 1-10 DOI: 10.1016/j.bandl.2007.08.003

Comments

Complex Brain Functions Are Not Reducible To Computer Logic

October 25, 2008 by johnbrandy, 1 year 4 weeks ago
Comment: 32367

RE: coglanglab, You wrote, "Does human behavior arise from the operations of the physical brain, or from the operations of a non-physical "soul" not bound by the laws of nature"? In my personal view, there are 'universal' forces, or principles that interact with the human brain that make consciousness possible. This is not a scientifically grounded view, as are not many of my statements. Much of my thinking is based largely upon direct observation, introspection, and insight. No doubt, I have 'dabbled' far and wide. Cognitive science is partially grounded upon; direct interviews, surveys, written test, and the like. These results are evaluated statistically, and compared with existing research in; cognitive science, psychology, neuroscience, and related fields of study. Doubtless, direct observation, introspection, and insight must play a necessary role in the recalling, organizing, and otherwise relating of solicited information from test participants. In other words, direct observation, introspection, and insight play a legitimate role in cognitive science, and perhaps lend a measure of veracity and fidelity to my remarks.

The possibility of universal forces or principles, as well as, "a non-physical soul", facilitating the modeling of the human brain, for the purpose of writing a valid computer program replicating any brain function, is not made clear. A separate agency, influencing and operating from outside the working brain, or, a "non-physical soul", directly linked to every thought, action, and deed, greatly complicates the defining and codifying of the relative and requisite elements necessary to create a working computer program capable of, among other operations, the formations of metaphors.

You stated, "If you agree that human behavior arises from the brain, which is governed by the laws of nature, it *must* be possible to model the brain -- and its behavior -- with a computer". The 'conclusion', "it *must* be possible to model the brain -- and its behavior -- with a computer", does not logically follow from the premise, "If you agree that human behavior arises from the brain, which is governed by the laws of nature". Until that potential connection is well established and reasoned out, I cannot agree with your conclusion.

You stated, "As far as the latter part of your comment, I believe it is a standard assumption in cognitive science that human behavior can be modeled by computers. The question is how, and what kind of computer". The assumption, opinion, or belief, that "human behavior can be modeled by computers", is not; as is absolutely necessary, the result of a sufficient understanding of brain structure and function, such that, any science can replicate, or even theorize, to a satisfactory degree, in pragmatic or operational terms, a computer model or program capable of such. I would suggest that you have effectively admitted to the fact that questions and issues relating to cognitive science, and by extension brain science are provisional, and approximations. Thus, the present difficulty, if not the impossibility of characterizing brain structure and function, in terms of computational language, capable of leading to a computer program that can replicate complex cognitive functions, such as complex metaphors, is predicated upon hope, and wishful thinking. And the belief that science can and will eventually answer all of the questions and issues about brain structure and function, and interpret such in compatible computational language and programs. Our present understanding of the above issues are, at best, speculative and theoretical.

The test of a theory

October 11, 2008 by coglanglab, 1 year 6 weeks ago
Comment: 32350

John:

You wrote, "It is not clear to me how a computer program could be created to solve complex metaphors. That possibility would imply a mathematical or computer program like understanding or interpretation of metaphors. I am not aware of such understanding or interpretation. Specifically, no such understanding or interpretation has been offered, or justified."

If you think this is so, I have a question for you:

Does human behavior arise from the operations of the physical brain, or from the operations of a non-physical "soul" not bound by the laws of nature?

If you agree that human behavior arises from the brain, which is governed by the laws of nature, it *must* be possible to model the brain -- and its behavior -- with a computer. Not necessarily any extant computer -- our technology has not yet reached that level -- but it must be at least in theory possible to build such a computer.

If not, then how are our brains built?

As far as the latter part of your comment, I believe it is a standard assumption in cognitive science that human behavior can be modeled by computers. The question is how, and what kind of computer.

---
Please try my web-based experiments

Deeper Explanations

October 18, 2008 by johnbrandy, 1 year 5 weeks ago
Comment: 32324

RE: coglanglab. I find your response interesting and useful. I think that a metaphor could be viewed as a special or particular kind of restatement. I find it personally intriguing you offered this judgment; though perhaps applied to my "explanation".

Restating a problem may very well clarify the issues, and, if not lead to a "solution", may represent "another step in that direction". In fact, restating the problem is probably the "most important" step in the process of arriving at a solution to questions and problems presented at a certain level of sophistication.

Certainly, in many mathematical problems, the solution represents a logical and systematic restatement of the initial problem. Methods for solving similar mathematical problems can be derived and deduced from the sequence of "elements" that constitute the solution. I'm sure this process is applicable to non-mathematical problems.

My purpose was not to simply restate the questions about metaphors, as important as that is. In point of fact, much of what I wrote is not stated or implied in your article.

If, as seems apparent, "Drawing abstract connections between largely unrelated objects to produce and understanding metaphorical language", requires creative synthesis, the "solution" to the problem of metaphors lies therein, and should constitute the primary area of investigation.

The capacity to make creative connections, at a given levels of sophistication is not necessarily amenable to inspection. Then again, given the requisite grounding, certain individuals can often explain how they made the required connections to produce meaningful metaphors. Some metaphors are easy to produce or evaluate, therefore the level of sophistication of the metaphors, and individuals, ask to form or evaluate them, must be figured in research designed to explain them. I see no argument here.

It is not clear to me; the rational for the suggestion that a computer program can be created to solve complex metaphors. That possibility would imply a mathematical or programmatically like understanding of metaphors, adequate to write a program capable of characterizing metaphors. I am not aware of such understanding, or theories that justify this possibility, or approach. It would be the same, in my opinion, as creating a program to reproduce creative synthesis.

The "how you know what connections to make", is not objectively discernible, with respect to certain complex metaphors. In fact, at certain levels of complexity, the abstract connections are not, so much "made", but the result, or product of unconsciousness thought associations. How does one describe, in detail, a creative process? The question as to "how you know what connections to make", ignores the unconscious element necessary to form certain metaphors, and their abstract connections.

Deep explanations

October 8, 2008 by coglanglab, 1 year 7 weeks ago
Comment: 32311

John:
I read your original comment as one offering a solution and replied to it in that vein. For instance, you wrote:

"In order for metaphors or analogies to take on meaning, they must, of necessity, relate to logical and reasoned connections, once established, meaningfully refer back to the original proposition or idea."

This is probably true, and I suspect that nearly everybody who has studied the question agrees on that point. The question is how you know which connections to make. Ultimately, until we have a computer model that "understands" novel metaphors and analogies, we don't have a complete theory. One example of a project that pushes that direction is work by Deidre Gentner.

Should my own writing be held to the same standard? Definitely. But the standard is not one of a complete explanation: there is no area in cognitive science where we are close to a complete explanation (low- and mid-level vision probably gets the closest). The goal, then, is to offer not a complete explanation, but an explanation that is more than just a restatement of the problem.

---
Please try my web-based experiments

This Is My Explanation

October 10, 2008 by johnbrandy, 1 year 6 weeks ago
Comment: 32281

My previous post is a rather detailed explanation, making note of the many elements and cognitive processes, or techniques that are necessary to "easily draw abstract connections between largely unrelated objects to produce and understand metaphorical language". My opinion, or proposition is not materially different from many of your articles, the explanations of which would require exacting research and results. Would you be comfortable being held to the same standard of certainty and exactitude?

I stated that the mystery implied, as to the formation or creation of metaphors, and their "abstract connections", is explicable. The explanations are contained in my previous comments,(post). Obviously, these explanations do not rise to the degree of exactitude and verifiability you seem to require, and do not always provide, in some of your articles. I find this troubling. After all, my 'opinion' is proffered, not as concrete fact, but as a proposition.

Admittedly, I failed to couch my remarks with language that made absolutely clear I was speaking in propositional terms. Therein, may lie your need for an explanation that rises to a higher degree of exactitude or verifiability. The "facts" in my exposition reflects a genuine understanding, or if you like, a belief as to the issues entertained. Of course, in spite of my confidence, you, or anyone else can disagree with my convictions, relative to the issue of metaphors.

Critical analysis requires specificity. Rather that merely asking me to offer an explanation,(which I did), why did you not specifically critique, point by point, that which was offered by me, and "demonstrate", or otherwise logically spell out how I "failed" to make a reasonable argument. In the absences of this common sense, 'scientific', and systematic approach, your statement, "ok, explain it", appears self-serving, and convenient.

I stated, "the mystery implied is explicable, though veiled in the process of creative synthesis". This is a complex process, involving all of the for mentioned components necessary to consciously or unconsciously "draw abstract connections between largely unrelated objects to produce and understand metaphorical language". Clearly, the abstract connections come from that which is contained in the mind. The ability to make the necessary metaphorical connections, must, of necessity, draw upon that which is contained in the mind. Specific metaphors, and their abstract connections cannot be established without having "mastered" specific kinds of knowledge, word meaning, grammar, and so forth. Why should this be so difficult to perceive or understand? Please, kindly critique in a meaningful and logical manner, why and how my explanation is unreasonable.

What is a connection, anyway, in a cognitive or metaphorical sense, if not relevant and specific thought associations, perceived consciously, or unconsciously,and derived from a working memory; shaped by education, knowledge, philosophy, word understanding and usage, literary knowledge, and so forth. The implied mystery, in pragmatic terms, is the difficulty, or impossibility of sorting through, identifying and organizing, in a step-lock manner, all of the relevant cognitive connections; relevant and specific thought associations, that lead to metaphorical thinking.

In my view, the primary question is how the process of creative syntheses actually makes the connections necessary to form or create metaphors, and abstract connections. Again, specific grounding, as mentioned, must play a fundamental role. Certainly, there are other factor. I would characterize them, in general terms as mind-set; education, mental habits, motivation, and perhaps necessity. Creative synthesis is not fully amenable to analysis, therefore, the essential question for further investigation, with respect to metaphors, and their abstract connections, is the nature and characteristics of creative synthesis, as it applies to the formation and creation of metaphors, and their abstract connections. This is my "explanation".

ok, explain it

October 6, 2008 by coglanglab, 1 year 7 weeks ago
Comment: 32274

John:

What you have said is quite possibly true, but it leaves a lot of questions unanswered. A complete theory of metaphor would essentially be a computer program which, given any metaphor, would explain what it meant.

I think we are a very long way from that computer program.

Please try my web-based experiments

The Mystery Implied is Explicable

October 5, 2008 by johnbrandy, 1 year 7 weeks ago
Comment: 32205

"This still leaves us a long way from understanding how humans so easily draw abstract connections between largely unrelated objects to produce and understand metaphorical language. But it's another step in that direction.". "Drawing abstract connections between largely unrelated objects to produce and understanding metaphorical language"; is the produce of having mastered, to a sufficient degree, the relationships between the meaning of particular words, words patterns, established word usage, and certain philosophical, grammatical, and literary understanding, sufficient to consciously, or unconsciously connect, and/or otherwise combine or synthesize seeming unique paradigms. The metaphorical relationship is always logically and analogically consistent with the origional idea or concept. The mystery implied is explicable, though veiled in the process of creative synthesis. In order for metaphors or analogies to take on meaning, they must, of necessity, relate to logical and reasoned connections, once established, meaningfully refer back to the origional proposition or idea.



About us

Science Blog was started in August 2002. It lives, breathes and eats press releases from research organizations around the globe. Most of what you read here are press releases from the outfits named in the stories themselves. Got a news story you think belongs here? Let's talk. The other half of the equation is blog posts from readers like you. So if you have an interest in science, please register and join others like you in an ongoing, vibrant dialog about what makes the world tick. Meantime, please take a minute to read our Privacy Policy and Site Disclaimer.


Premium Drupal Themes by Adaptivethemes