A Connectionist Model of Metaphor

In my posts on metaphor (I, II, III, and IV), I focused primarily on theories of metaphor, with some empirical work thrown in for good measure. I didn't discuss any of the many models that implement these theories, primarily because there are so many. Today I was reading a paper on a connectionist model, called the Metaphor by Pattern Completion (MPC) model (Thomas & Mareschal, 1999), that implements the attributive categorization model (see posts III and IV), and I thought it might be interesting to give a quick description of it.

The model uses a pretty normal three-level connectionist architecture (see here for a good primer on connectionism), consisting of input and output nodes, along with intervening hidden nodes. The model was trained with exemplars from three separate categories (Apple, Fork, and Ball), and the representations of these three categories are stored in the hidden units. Here is a graphic representation of the architecture:



Figure 1 from Thomas and Mareschal (1999). Click for larger view.


Given an exemplar, the model autoassociates the features that comprise the representation of the category to which the exemplar belongs. This just means that the model reproduces the features, in the form of semantic vectors, of the category as output. So, given an exemplar with a set of features, the model will determine into which category the exemplar belongs, and output the features of that category. To model metaphors, an exemplar serving as the topic of the metaphor is inputted into the node(s) representing the category which serves as the vehicle. For instance, the metaphor "An apple is a ball" would involve inputting an apple exemplar into the node(s) representing the category Ball. The output, in this case, would be a representation of apples (the topic) which has been altered by the vehicle representation so that the output is more similar to the ball representation than the original exemplar was. Here is the author's description of why this alteration of the topic representation occurs:

Pattern completion is a property of connectionist networks that derives from their non-linear processing (Rumelhart & McClelland, 1986). A network trained to respond to a given input set will still respond adequately given noisy versions of the input patterns. For example, if an autoassociator is trained to reproduce the vector <0> and is subsequently given the input <.2 .6 .2 .2>, its output is likely to be much closer to the vector it 'knows', perhaps <.0 .9 .0 .0>. An input is transformed so as to make it more consistent with the knowledge that the network has been previously trained on. The connection weights store the feature correlation information in previously experienced examples. If a partial input is presented to the network, it can use that correlation information to reconstruct the missing features.

When the model is actually used, it exhibits properties of metaphors consistent with the attributive categorization theory. First, the representation of the topic and vehicle interact to determine which features of the topic are represented in the output of the metaphor. In addition, the output of the metaphorical comparison differs depending on the direction of the comparison. "An apple is a ball" will produce a representation that is different from "A ball is an apple," indicating that the model produces results consistent with the irreversability of metaphorical statements. In addition to the results consistent with the predictions of the attributive categorization theory, the model itself produces three new predictions. The first is that the smaller the range of features found in exemplars of a vehicle category, the less metaphorical comparisons involving that category as a vehicle will produce interactive effects. In other words, when a vehicle with a small range of features is used in metaphors, the features it transfers to the topic will tend to be the same regardless of what the topic is. This will also be the case when the vehicle category is highly familiar. Finally, metaphorical comparisons will involve the transfer of attributes from the vehicle to the topic that are not likely to be reported. This is because, while some features may not make sense when transferred from the vehicle to the topic, (e.g., "A ball is an apple" may transfer the feature "edible," even though very few balls are actually edible), but which are transferred anyway. As a result, Thomas & Mareschal predict that participants will be slower to questions about features which are transferred but not otherwise reported.

So, there you have it, a connectionist model of metaphor implementing the attributive categorization theory. There are several obvious problems and limitations with the model. For one, it's not really doing what the attributive categorization theory says is involved in metaphor. Recall that in the attributive categorization model, the vehicle is a member of a category. The topic is placed into that category, not into the category specifically referred to by the vehicle label. For example, "My job is a jail" doesn't involve categorizing my job as a jail, but as a member of a common category, confining plances. However, in the MPC model, the topic is placed into the category referred to by the vehicle label, rather than a common category. I imagine the model could be modified to do this, but as it's described, it isn't doing what the attributive categorization theory says it should be. In addition, the model can only deal with metaphors that involve the transfer of attributes, and cannot handle any metaphors that involve the transfer of structure. This is a big problem, since research has shown that most metaphors involve structural information. In addition, the model's version of the irreversability of metaphorical comparisons is pretty weak. The model can perform comparisons in either direction, and has no way of determining which direction is better. This hints at a limitation common to most (if not all) connectionist models: it's not clear how th model fits within a larger system which, in this case, would ultimately be needed to explain the full range of metaphorical behaviors we humans exhibit. The model's simulation of metaphor is therefore uncomfortably elliptical. Finally, with the exception of the third prediction, the first two predictions are hardly novel, and are pretty much self-evident. If the model had not demonstrated these properties, it would certainly have been in trouble, but the fact that it does make these predictions is hardly a case for taking the model seriously. The third prediction is interesting, but it is also a prediction that would be made by most comparison theories, and therefore wouldn't allow us to differentiate between this model and comparison models.

Buridan's Ass

At Oohlah's Blog-space, there are two recent posts (this one and that one) on "choice without preference" and Buridan's Ass. For those who are too lazy to click the links, here's the first of the two posts in its entirety:

In the first chapter of Jon Elster's Solomonic Judgments, he argues that choice without preference is not an important practical issue. He contends that no one cares which of two apparently identical soup cans on the supermarket shelf is chosen. The only way a choice like this will matter is if there are differences in the two soup cans, i.e., one has more broth than the other, etc.

I may have missed the point, but the problem of the choice without preference is that we can choose either soup can A or soup can B and satisfy our desire for buying soup. Since both soup cans will satisfy our desire, then rationality tells us to purchase both soup cans. If we think like this, though, we could potentially become poor very quickly. (Perhaps using the example of very similar cars on a new or used car lot would bring the problem to the fore.) So, it is not rational to purchase both cans. What seems to follow is that reason tells us to do something it is not rational to do.

Have I missed the point of choice without preference, or has Elster missed an important component of this difficult problem?

I may be stupid, but I'm like Elster -- I just don't see a problem here. The scenario assumes that our only motivation is to purchase the can, and that there are no constraints other than preference on choice. It ignores the motivation the post mentioned (the financial motivation to only buy one can), time-constraints, motor constraints, etc., which all factor in to the optimal (which is really what "rational" is in this case) choice. If you get rid of all of those, then there's probably no need to choose a can in the first place. In fact, I bet that if you threw all of those into an Ideal Observer model, what you'd find is that the model either chooses one of the two cans all of the time (which would probably be due to the motor constraints) or choose each can 50% of the time (because the motor constraint doesn't apply). Humans, because we're not ideal observers (our behavior is suboptimal) will have all sorts of noise in our data if forced to make the choice a bunch of times, but it would probably be pretty close to the ideal observer model assuming that the initial conditions were the same every time (which would be impossible, but you get the point).

Maybe Buridan's Ass is an interesting logical problem, even though it's not an interesting practical problem, and probably can't shed any light on the decision-making process (though it apparently sheds some light on Romance novels -- see here) other than that we have reasons for making decisions that go beyond the independent attractiveness of two options (duh!), but I can't imagine how it would be. For it to be even a logical problem would require that we pretty much remove all of the motivations to make the choice in the first place (a similar point for is made by Richard at Philosophy, etcetera). If it were still a problem, logical or otherwise, even given all of the constraints and other motivations that factor into a choice like this (or with them all removed), here's how I would solve it: I'd pick up both cans and then make a choice when I got to the cashier. By that time, choosing one should be more optimal than choosing the other (maybe because it's closer to me in the cart). I think that's pretty much the solution people much smarter than me have traditionally come up with (e.g., the solution of Otto Neurath, which just involved flipping a coin). As for the ass, since he probably doesn't have a shopping cart, or a coin, he's just going to have to suck it up and make a choice.

A Plea for More Requests

For those of you who find my political rants irksome, or worse, I just wanted to let you know that two cognition posts or in the offing, one on creative cognition and one on the cognitive psychology of humor (by request). I also want to ask (nay, beg) for more requests. In the post-election blog atmosphere, I'm having trouble coming up with ideas for posts. If there's anything you'd like to read about, let me know. (Also, I haven't forgotten the connectionism request, but I'm writing something on representations right now, and I may either post a short version of that, or summarize the ideas in it, when I'm finished.)

The Conservatives We Don't Want

The Anal"Philosopher" writes:

By allowing Muslims to immigrate into their countries in massive numbers, Europeans are destroying their culture and risking their lives. What idiots. And these are the people we are supposed to emulate with regard to gun ownership, capital punishment, homosexual "marriage," confiscatory taxation, and other matters? Americans must never stoop to the level of Europeans. If they do, they deserve Europe's fate.

I don't know about you, but to my ear, that sounds exactly like the arguments heard in the United States during the first few decades of the twentieth century, but instead of Muslims, the arguments were directed at Italians and the Irish. It also sounds uncannily like the arguments made in the south about the effects of freeing the slaves and loosing all those dark-skinned folk on the fair, southern culture. These are the Republican voters we don't want to reach out to, and should ignore entirely. This may sound harsh, but what I think what they want is less than worthless.

Trying to Reach the Evangelicals, On Purpose

Apparently it's fashionable among liberal bloggers (and non-bloggers) to wonder out loud about how Democrats can court the evangelical vote. For instance, Lindsay Beyerstein of Majikthise writes on the topic, concluding:

I don't know whether Democrats can find equally compelling areas of common ground with evangelical groups today. Perhaps Democrats can make common cause with evangelical groups over economic justice, peace, or other issues of mutual concern. The alliance between the religious right and the corporate right may seem inevitable in retrospect, but it too is built on a sometimes uneasy compromise. Proponents of religious alternative outreach are urging Democrats to cultivate compromise in the other direction (economic interests over cultural interests). Whether such outreach is feasible remains an empirical question.

While Lindsay seems to harbor at least some hope of someday getting more evangelical votes, I do not, and to be honest, I don't want them. The evangelicals who don't already vote Democrat (and there are some; my parents make at least two) aren't the sort who are swayed by economic policies. The ones who are concerned with economic policy are likely to harbor a conservative, hyper-individualist, free market view of economic issues, and any talk of something like universal healthcare or government-funded arts or environmental programs will turn them off immediately. If we talked about gun-control, prison reform, or progressive taxation, we might cause these evangelicals to have an aneurysm. Then there are those who don't vote Democrat for moral and religious reasons. These are the people who vote based on issues like abortion, same-sex marriage, sex education, evolution, and the like. There simply is no reaching these two classes of evangelicals without compromising our core values, and as Lindsay herself says, "Compromise outreach to is morally unacceptable and politically naive."

Instead of focusing on evangelicals, Democrats should be focusing on the people who do sometimes vote Democratic. Some of these voters may be evangelicals, and they may even hold some of the views that the untouchable evangelicals hold, but they don't vote on these issues alone. The way to get to them is to make it clear that Democrats will do more for them, in the short-term and the long-term, than Republicans will. People like me seem to think this is obvious, but the swing voters, evangelical or not, don't see it. So, it's not as obvious as we think, and we liberals got to do a better job of expressing ourselves and getting the message out there.

My one true hope, and I think this is a genuine possibility, is that the Democratic party will become more liberal, not less so, by recognizing that the conservative moral and economic voters are out of their reach, and focusing on what differentiates us from them. When survey after survey shows that a substantial majority of Americans support universal health care, why do we de-emphasize this issue in elections? Because we think we will lose the ultra-conservative voters? We've already lost them. The more we talk about trying to win them, the further we get from what we should be talking about.

Philosophers' Carnival IV: Attack of the Epistemologists

The fifth Philosophers' Carnival is up at Ciceronian Review. There are some interesting posts this time around. The post from Prosthesis on science and beauty is very interesting, and reminds me of Stephen J. Gould's work towards the end of his life (leading to this book, and research at this lab). It's hard to tell whether the author of the post at Prosthesis thinks "nonempirical" factors in theory selection are bad for science, but it's clear that he/she thinks they are unavoidable. I like Gould's take, in which nonempirical (most notably aesthetic) factors are not only inevitable, but desirable. Science is about understanding the world around us, after all, and aesthetic considerations can facilitate understanding.

Blowhard has a very good introduction to Stephen Toulmin, whom I only recently really discovered, so I found the post helpful.

Siris, which continues to be one of my favorite blogs, has a good post on Shepherd on causation. I don't really know anything about Shepherd, and I've only really begun to think seriously about causation, so this post was definitely edifying for me.

I also liked the post at Evolving Thoughts on knowledge of the past in science. The issues discussed in the post don't really come up in my own work as problems with scientific knowledge, but similar issues are raised when trying to verify the accuracy of memories. For instance, in the debate over recovered memories, what evidence constitutes verification of the accuracy of the content of recovered memories? Unlike in the sciences, immediate certainty is important here, because the future discovery of new evidence may come too late. For instance, when recovered memories are used as the primary evidence in criminal cases against accused child molestors, we can't wait until after the trial to decide whether the recovered memories are accurate.

There are several other interesting posts in this edition of the Carnival as well, so go check them out.

Concepts of "Human" Among American Undergraduates

I've been reading a lot of literature on the creative cognition approach lately (expect a post soon), and came across something interesting, though it's not really about creative cognition. In an experiment by Ward et al., participants (98 undergraduates in an introductory psychology course) were asked to list properties of humans that distinguish them from animals. Here are the most common categories of properties listed, along with the percentage of participants who listed them (taken from Table 1 in Ward, et al., p.1390):

Communication 72
Mental ability 68
Physical features 42
Technical/manipulative 33
Emotions 32
Socioeconomic institutions 25
Morality/religion 22
Sex/reproduction 20
Clothing 13
Instinct 15
Family 12
Lifespan 11
Consciousness 8
Dominance 7
Creativity 6

How do their concepts of "human" fit with yours? I think it's interesting that "consciousness" was only listed by 8% of the participants, when it's played such a large role in intellectual debates about "humanness."

Cognitive Science and Literary Criticism

A while back I discovered a very interesting, and perhaps even important (though, at least in the cognitive sciences, almost completely unread) paper by Herbert Simon, one of the few psychologists to have won the Nobel Prize. The paper, which can be found here along with several commentaries, attempts to "bridge the gap" between cognitive science and literay criticism. The essay focuses primarily on meaning, because literary criticism is a practice primarily aimed at analyzing the meanings of texts. His account of meaning is clearly impoverished. It revolves around memory retrieval. Here are two paragraphs from the paper that present the gist of what Simon has to say about meaning:

Meanings are evoked. When a reader attends to words in a text, certain symbols or symbol structures that are stored in that reader's memory come to awareness. (In psychology we might say, more ponderously, "having been noticed, the symbols are activated or transferred from long-term to short-term or immediate memory"). This is the sense in which we will use the term "evoke" throughout this paper. It denotes a specific set of psychological processes that have been much studied in the laboratory and in everyday life: the processes that bring meanings, or components of meaning, into attention.

The process that underlies evocation is recognition. Words in the text serve as cues. Being familiar (if they are not familiar, they will not convey meaning), they are recognized, and the act of recognition gives access to some of the information that has been stored in association with them--their meaning (Feigenbaum and Simon, 1984). Recognizing a word has the same effect as recognizing anything else (a friend on the street). Recognition accesses meaning.

Then, a little later, he adds:

The meaning of the text, then, will be a function of the memory contents that are accessed by recognition of words. Which of the whole collection of memory contents will be accessed depends on context, that is upon what contents are both associated directly or indirectly with the word recognized and also the extent to which they have recently been activated.

And to complete the summary:

Thus, the evocation of certain symbols may evoke others by the chain reaction that we call mental association. The burst of evocation that a bit of text may induce is limited only by the richness and complexity of the memory structures that it activates. The more elaborate the structures that are evoked, the more the meaning to the reader is defined by the reader's memory, the less by the author's words. The meaning of text is determined by a relation between the text itself and the current state of the memory of the reader, its contents, and its state of activation. (emphasis added)

I agree with Simon that literary criticism can benefit from a deeper connection with cognitive science (though I'm not so convinced the relationship can be symmetrical), but I also agree with some of the commentators (this one and this one, for istance) that Simon's conception of meaning isn't going to to anyone, much less literary critics, much good. If it were only that his description of the cognitive scientific view of meaning was an oversimplification, I think that would be OK. It's a short paper, and cognitive science has a lot to say about memory. However, I think oversimplicity is not the only problem. The paper's account of meaning is just wrong. It's wrong in how it describes memory (which is probably more case-based, more reconstructive, and much less encyclopedic than he would lead us to believe), and I think he pays far too little attention (none, in most cases) to things like inference, imagery, layers of meaning, and the role of creativity in extending meaning. Still, with all its flaws, I think the aim of the paper is a good one.

I know what you're thinking: how does Chris think cognitive science can benefit literary criticism? Excellent question. The most obvious way is by developing an understanding of the cognitive processes involved in reading. In addition, the elucidation of the cognitive processes underlying literary techniques, like metaphor, imagery, and the construction of representations can benefit the study of texts. Finally, cognitive scientists can help critics to understand creative cognition. How are novel representations produced? How do we create entirely fictional worlds out of the material of the factual world? None of these things is designed to give literary critics a single, all-encompassing view of meaning with which they can interpret all texts. There may not be such a theory, and if there is, cognitive science isn't yet equipped to give it. Still, each of those things plays a role in meaning construction, in the minds of both authors and readers.

Anyway, I recommend checking out the paper, and the commentaries, if this is the sort of thing that interests you. Simon is not much of a writer, but he's very insightful, and even when he's wrong, he still raises important issues.

Promiscuous Females = Better Semen

I'm not really a big fan of evolutionary psychology. I don't think it has anything important to say about cognition, and at least up to this point, I've been right. However, there are interesting ways to use evolution to study things like sexual behavior. You have to do good research, though. For instance, this is the sort of research that evolutionary psychologists should be doing, rather than this shit. Here's an excerpt from the article on the study (the good one):

Researchers have shown that when females are more promiscuous, males have to work harder -- at the genetic level, that is. More specifically, they determined that a protein controlling semen viscosity evolves more rapidly in primate species with promiscuous females than in monogamous species. The finding demonstrates that sexual competition among males is evident at the molecular level, as well as at behavioral and physiological levels...
Lahn's group studied semenogelin, a major protein in the seminal fluid that controls the viscosity of semen immediately following ejaculation. In some species of primates, it allows semen to remain quite liquid after ejaculation, but in others, semenogelin molecules chemically crosslink with one another, increasing the viscosity of semen. In some extreme cases, semenogelin's effects on viscosity are so strong that the semen becomes a solid plug in the vagina. According to Lahn, such plugs might serve as a sort of molecular "chastity belt" to prevent fertilization by the sperm of subsequent suitors, though they might also prevent semen backflow to increase the likelihood of fertilization.

Lahn and his colleagues compared the SEMG2 gene, which contains the blueprint for semenogelin, from a variety of primates. They began by sequencing the SEMG2 gene in humans, chimpanzees, pygmy chimpanzees, gorillas, orangutans, gibbons, macaques, colobus monkeys, and spider monkeys. These species were chosen because they represent all the major mating systems, including those in which one female copulates with one male in a fertile period (such as gorillas and gibbons); those in which females copulate highly promiscuously (such as chimpanzees and macaques); and those in which mating practices fall somewhere in between (such as orangutans where a female will copulate with the dominant male, but may also copulate with other males opportunistically).

"When we plotted data on the evolution rate of the semenogelin protein against the level of female promiscuity, we saw a clear correlation whereby species with more promiscuous females showed much higher rates of protein evolution than species with more monogamous females," said Lahn. The researchers measured protein evolution rates by counting the number of amino acid changes in the protein, then scaling it to the amount of evolutionary time taken to make those changes.

"The idea is that in species with promiscuous females, there's more selective pressure for the male to make his semen more competitive. It's similar to the pressures of a competitive marketplace. In such a marketplace, competitors have to constantly change their products to make them better, to give them an edge over their rivals -- whereas, in a monopoly, there's no incentive to change."

Now this is the sort of research I can respect. Of course, it's done by biologists, not psychologists, which may explain why it's good evolutionary research.

Deaths in Iraq

There's been a lot of really, really bad commentary on the Lancet study in the blogosphere. Apparently everyone becomes a self-described expert in statistics when the statistics disagree with your previously held beliefs. At least we have Timothy Lambert and Daniel Davies to set the record straight. Lambert does an excellent job (as usual) of debunking some of the most common criticisms, and links to other defenses of the study, while Davies provides a thorough look at other common criticisms.

I really can't help but wonder why so many people who've clearly never taken a statistics course, and some who obviously haven't even read the paper, feel qualified to comment on its methodology and findings. Even those who do have some statistical knowledge (or at least should), seem to forget everything they've learned (or should have learned) when commenting on the paper. Thus, John Ray, who has published some admittedly unimportant psychological research that used statistics, writes:

Nobody, however, seems to have commented on the fact that the findings were a product of cluster sampling. The major fault I see with the study is that estimating low-incidence phenomena via cluster samples is inherently dodgy. I have had many findings derived from cluster samples reported in the academic journals so I know a little bit about it. You just have to get one or two clusters being a-typical (either by chance or intentionally) to arrive at totally distorted results. Basing such an important conclusion on a sample-size of only 33 is really quite ludicrous. I have used as few as 10 clusters in some of my surveys but I was concerned only to find whether some effect existed at all. I was not trying to estimate it precisely.

I'm going to be charitable and assume that it's been so long since Ray has published anything using cluster sampling that he doesn't remember how it works, and what sort of estimation errors one is likely to get using it. If that's the case (instead of Ray being so upset that the study disagrees with his personal beliefs about Iraq that he is rendered "temporarily" stupid), I hope that Ray and others will read Davies' post, because he explains where Ray goes wrong. He writes:

Although sampling textbooks warn against the cluster methodology in cases like this, they are very clear about the fact that the reason why it is risky is that it carries a very significant danger of underestimating the rare effects, not overestimating them.

Hopefully, further research will be done on the effects of the war in Iraq on Iraqis. This study is really just a preliminary one, though its findings should be taken seriously. However, as the results of future studies are likely to further upset conservatives, I can't imagine their responses will be any different, no matter how many studies demonstrate that Iraqis are worse, rather than better off, in post-war Iraq.

One More on the Great War

Since the anniversary of the Treaty of Versailles is now Veterans Day, which seems appropriate since it is this treaty that led to the creation of millions of future veterans of foriegn wars, Lindsay Beyerstein gives us "In Flanders Field," a poem written in 1915, during the Second Battle of Ypres, before McCrae could have known just how many more would die. Instead offering a another poem that glorifies the dead directly, I think I'll post one with a different perspective on the men who fought, and society in general. I think it does a good job of capturing the war, and the atmosphere that allowed it. Sometimes this sort of commentary can do more to honor those who died than direct tributes. I also think the poem is disturbingly appropriate for today. The entire poem is here. I will only give you a piece:

From Ode Pour L'election De Son Sepulchre by Ezra Pound

IV
These fought in any case,
And some believing,
pro domo, in any case...

Some quick to arm,
some for adventure,
some from fear of weakness,
some from fear of censure,
some for love of slaughter, in imagination,
learning later...
some in fear, learning love of slaughter;

Died some, pro patria,
non "dulce" not "et decor"...
walked eye-deep in hell
believing old men's lies, then unbelieving
came home, home to a lie,
home to many deceits,
home to old lies and new infamy;
usury age-old and age-thick
and liars in public places.

Daring as never before, wastage as never before.
Young blood and high blood,
fair cheeks, and fine bodies;

fortitude as never before

frankness as never before,
disillusions as never told in the old days,
hysterias, trench confessions,
laughter out of dead bellies.

V
There died a myriad,
And of the best, among them,
For an old bitch gone in the teeth,
For a botched civilization,

Charm, smiling at the good mouth,
Quick eyes gone under earth's lid,

For two gross of broken statues,
For a few thousand battered books.

The Anniversary of the Treaty of Versailles

John Quiggin has an excellent post on the anniversary of the treaty that ended The War to End All Wars. This war more than any other of the last century, and perhaps since the Greeks themselves, seems to embody the essence of tragedy, with the folly, shortsightedness, greed, ineptitude, and all around ineluctable humanness displayed throughout it. The Treaty of Versailles, as ill-conceived as it looks to us now in hindsight, is merely another event born of the same tragic impulses that drove virtually every other act of the war. Yet though it is often almost unbearable to consider the mistakes and their later implications, much less the millions of dead, the Great War has always fascinated me to no end. I've spent countless hours reading about it or looking at pictures, imagining what life must have been like for those ordinary soldiers who knew full well that their chances of being killed and wounded were greater than their chances of making it out unscathed (counting those taken prisoner, the casualty rates for the French, Russians, Germans, and Romanians were all over 60%, while 90% of the Austria-Hungarians who fought in the war were killed, wounded, or captured). In the end, all I can do in the face of such mindless destruction is echo Quiggin's final sentiments:

War is among the greatest of crimes. It may be the lesser evil on rare occasions, but it is always a crime.

It's a shame so few have learned any real lessons from that war, or any other. They certainly haven't learned the lesson Quiggin has.

Note: On this site you will find some amazing color (not colorized!) photos from World War I. I highly recommend it if you are interested in that sort of thing.

Counterfactuals and the Real World

Brandon has a great post on counterfactuals at Siris, in which he argues that many counterfactuals are designed to highlight something factual. He gives some good examples, such as, "If I were a Hindu, I would worship Shiva," and analyzes them in order to show what it is they are meant to show. Here is his analysis of that counterfactual:

Suppose I said something like this. Not only am I not a Hindu, I can directly connect my preference for Shiva with a number of things in my personal history that would not at all have been likely to occur had I actually ever been a Hindu. What (1) really imports is not something about what I would od if I were actually a Hindu; its import is that I (now, as I am) see reasons for thinking a certain thing about Shiva-worship (what that thing is will depend on the context; it might be a comment implying that I think Shiva-worship is preferable to other sorts of Hindu worship, or that I think it more intelligible, or that I think it more in line with my temperament, or some such). In other words, although put idiomatically in a counterfactual form, what it really is set out to describe is something factual.

I think he's right about the examples he gives, and would go even further to say that all counterfactuals serve to highlight aspects of the factual world. Take, for example, counterfactuals used in causal reasoning. Imagine that I have just been in a car accident, and I am thinking about what I could have done to avoid it. I might think that if I had taken my usual route, rather than the scenic one, I would not have gotten into the accident. Or I might wonder if I had swerved instead of hitting my breaks, I would have been able to miss the car that I ran into. In these cases, the counterfactual serves two purposes. The first is to come up with possible scenarios in which I would not have gotten into an accident. This then serves to help me to understand what aspects of the factual scenario actually caused the accident, which is the second, and in most cases primary goal of the counterfactual.

In addition to being used in causal reasoning, counterfactuals are also often used for rhetorical purposes. For instance, Gilles Fauconnier has analyzed the counterfactual, "If [Bill] Clinton had been the Titanic, the iceberg would have sunk." This is an excellent example of a figurative counterfactual designed to highlight something about the factual Bill Clinton. This counterfactual was used (I forget by whom) soon after Clinton was impeached and it became clear that congressional Republicans were hurt more by the impeachment process than Clinton himself. The counterfactual is meant to highlight these aspects of the factual world. It essentially says that no matter what you bring against Clinton, you, rather than he, will ultimately be the one who gets hurt.

The fact that counterfactuals are generally used to highlight some aspect of reality demonstrates the importance of the mapping process in the production and comprehension of counterfactuals. As I've said previously, mappings between the factual and counterfactual scenarios serve to structure the counterfactual scenarios using our representations of the factual ones. In addition to providing structure for the counterfactual scenario, though, the mapping can also serve to highlight aspects of the factual scenario through contrasting alignable differences in the two domains. In other words, elements of the factual and counterfactual scenarios that are mapped onto each other (because they play the same role in the relational structure shared by the two scenarios), but which differ in some way, are brought to our attention through the mapping process. This highlighting of alignable differences has been demonstrated in research on analogy and metaphor, and is one of the key features in the structural alignment models of those phenomena. The fact that highlighting alignable differences through the mapping process can help explain the fact that counterfactuals serve to call our attention to certain aspects of the factual world makes me more confident that this sort of model can be used to explain counterfactuals as well.

Idioms, Metaphors, and Lakoff, Oh My!

Now that the election is over, it's safe to talk about Lakoff and his theory of metaphor, but before I get to that, I want to talk about idioms. Idiomatic expressions are interesting because in many cases the connection between them and their meaning is not always obvious. Take the idiom "kicked the bucket." What does kicking the bucket have to do with death? There are all sorts of folk etymologies constructed for these sorts of idioms (e.g., kicking a bucket on which one stands to hang oneself), but for the most part, the real phrase-meaning connection remains elusive¹. For practical purposes, it might seem as though these connections don't matter; convention has established a meaning for idiomatic expressions, and people are able to learn them even when the expressions are opaque, as is "kicked the bucket." However, it turns out that the tendency to search for the connections between idiomatic expressions and their conventional meanings is in fact imporant for both practical and theoretical reasons. To demonstrate why, I'll quickly describe an experiment on idiom comprehension. Afterwards, I'll talk a little about the implications of this experiment, and some others, for Lakoff's conceptual metaphor theory.

Keysar and Bly ² gave participants unfamiliar idioms³ in contexts that implied one of two meanings for the idioms. Half of the participants read the idiom with one meaning, and the other half with another. Afterwards, participants were asked to rate how likely it was that the idioms could have another meaning. Keysar and Bly found that after exposure to one meaning, participants; ratings of the likelihood that the expression could have another meaning were much lower (than another set of participants who read the idioms without being given a meaning). This effect grew stronger the more participants were exposed to the first meaning of the idiom. Furthermore, participants spontaneously constructed explanations for the meanings of the idioms. For instance, when given the idiom "the goose hangs high" in a context in which it meant that things were going well, participants might interpret it as meaning "there is a freshly-killed goose hanging in the larder, and so there will be plenty of food." When asked how likely it was that "the goose hangs high" meant things were not going well, participants who had been given the "things are going well" meaning rated this as very unlikely. Thus it appears that because people assume that there is a connection between the expression and its conventional meaning (and even construct explanations for this connection), and that people have a difficult time believing that the idiom could have another meaning once they've given it an interpretation.

While plausible, peoples' inferences about the literal meanings of the idiomatic expressions in the Keysar and Bly experiment were not based on any real evidence. Instead, they were "best guesses" based on the expression itself and the meaning it was given. One of the motivations for this experiment was to argue that this is the sort of thing that appears to be going on in cognitive linguistics when people like Lakoff and Johnson interpret conventional expressions such as those that use the language of war to talk about arguments (e.g., "The debate teams battled hard"). According to Lakoff and Johnson, such expressions are instantiations of conceptual metaphors (in this case, the "argument is war" metaphor). When people interpret these conventional expressions, they are making conceptual mappings between the domain being discussed (e.g., arguments) and a base domain (e.g., war). Under this view, this how we understand these expressions each time we hear them (i.e., we have to make the mappings for the expressions to make sense). Keysar and Bly argue that at most, these interpretations (specifically Lakoff and Johnson's interpretations) of conventional expressions are post hoc inferences like those the participants made about the meaning of "the goose hangs high." Instead of making the conceptual mappings implied by Lakoff and Johnson's intepretations of such statements, people (including Lakoff and Johnson) may build these interpretations after comprehension. The metaphorical mapping between arguments and wars is not actually part of the meaning of the expression itself, but merely a spontaneous explanation of that meaning. Here's a further explanation of the differences between the two views, from Keysar et al.⁴:

To point out the difference between the two alternatives, consider Lakoff and Johnson’s claim that “it is important to see that we don’t just talk about arguments in terms of war. We can actually win or lose arguments [. . .] It is in this sense that the ARGUMENT IS WAR metaphor is one that we live by in this culture; it structures the actions we perform in arguing.” Our alternative claim is that we usually do “just talk” about arguments using terms that are also used to talk about war. Put more simply, the words that we use to talk about war and to talk about arguments are polysemous, but systematically related. Just as a word such as depress can be used to talk about either physical depression or emotional depression, words such as win or lose can be used to talk about arguments, wars, gambling, and romances, with no necessary implication that any one of these domains provides the conceptual underpinning for any or all of the others. The bottom line is that conventional expressions can be understood directly, without recourse to underlying conceptual mappings. Thus, when we say that an argument is right on target we do “just talk” about arguments using terms that we also happen to use when we talk about war—and music, art, literature, journalism, film criticism, and any other human activity in which something can be more or less on target. (p. 578)

Things get still worse for Lakoff and Johnson when we consider more empirical evidence. The fact that people spontaneously produce post hoc interpretations of the meanings of conventional expressions calls into question Lakoff and Johnson's own metaphorical interpretations of such expressions, but it doesn't show that people aren't actually making the conceptual mappings Lakoff and Johnson say they are. However, other findings make it clear that people really aren't making these sorts of conceptual mappings. For instance, in one set of experiments, McGlone⁵ asked participants to paraphrase conventional metaphorical expressions like those about arguments that use terminology from the war domain. Participants produced paraphrases that were consistent with the meaning of the expression (e.g., a long lively argument), but rarely produced paraphrases that said anything about the base domain (e.g., talk about war), implying that no mappings had occurred.

In another set of experiments, Keysar et al. had people read scenarios that contained either no mapping, an implicit mapping (i.e., they used language that Lakoff and Johnson argue involves the instatiation of a conceptual metaphor, but the metaphor itself was not made explicit in the sentences), or an explicit mapping (same as in the implicit condition, but with the metaphor itself as one of the sentences in the scenario). Example scenarios from the "love is a patient" metaphor are below (from Keysar et al., Table 1):

No mapping “Love is a challenge” said Lisa. “I feel that this relationship is in trouble. How can we have an enduring marriage if you keep admiring other women?” “It’s your jealousy,” said Tom.

Implicit “Love is a challenge” said Lisa. “I feel that this relationship is on its last legs. How can we have a strong marriage if you keep admiring other women?” “It’s your jealousy,” said Tom.

Explicit “Love is a patient,” said Lisa. “I feel that this relationship is on its last legs. How can we have a strong marriage if you keep admiring other women?” “It’s your jealousy,” said Tom.

At the end of each scenario was the same target sentence, which referenced the conceptual mapping. The target sentence for the "love is a patient" scenarios was, "You're infected with this disease." Keysar et al. measured reading times for this target sentence, and compared them across conditions. If conceptual mappings occur during the comprehension of the implicit or explicit mapping scenarios, then it should be easier to comprehend the mapping-related target sentence, and therefore reading times for these sentences would be shorter in the two mapping conditions (the literal scenario was included as a manipulation check). If people don't spontaneously conduct the mapping, but do so when prompted to, then the difference in reading times should only show up in the explicit mapping scenarios. If people aren't making mappings, then there should be no difference between the no-mapping and mapping scenarios. This last possibility is what Keysar et al. actually found. Reading times did not differ across the three conditions. Thus it appears that people weren't making the mappings when interpreting statements that should, according to Lakoff and Johnson, require conceptual mappings in order to make sense. They didn't even make the mappings when they were prompted to by sentences making the mappings explicit. In fact, the only time in which they did appear to make the mappings Lakoff and Johnson say they should occurred in another experiment, in which novel (rather than conventional) metaphorical statements were used. For example, the following is a novel scenario derived from the "love is a patient" metaphor (from Table 1):

Novel “Love is a patient,” said Lisa. “I feel that this relationship is about to flatline. How can we administer the right medicine if you keep admiring other women?” “It’s your jealousy,” said Tom.

When given this scenario, participants did read the target sentence faster than in the no-mapping condition, as well as the implicit and explicit mapping condition. Thus, it appears that novel metaphors do require mappings, while conventional expressions do not.

Where does all of this leave conceptual metaphor theory? Well, I don't know about you, but I think it's pretty safe to start treating it as a result of the illusory transparency of conventional expressions, rather than as a good theory of everyday thinking. The evidence overwhelmingly indicates that people simply aren't performing the conceptual mappings that the Lakoff and Johnson conceptual metaphor theory requires. Fortunately, outside of the cognitive linguistics circle, this is how Lakoff and Johnson's theory is already viewed. However, Lakoff has bipassed the cognitive science world, and taken his theory straight to the public in the form of his framing analysis of political discourse. This is unfortunate. If we try to do framing the way Lakoff tells us we should, we're going to quickly run into problems. Lakoff's entire analysis of the conceptual metaphors underlying the two poles in American politics is probably nothing more than a misguided (and painfully bad) attempt to explain the meaning of "the goose hangs high." If we try to use these conceptual metaphors to explain and sell the moral underpinnings of our political views, we're simply not going to activate the desired mappings, as the Keysar et al. experiments show.

¹ I've heard that "kicked the bucket" has its origins in pig-slaughtering techniques, but I'm not sure how reliable that etymology is either.
Keysar, K, & Bly, B. (1995). Intuitions of the transparency of idioms: Can one keep a secret by spilling the beans? Journal of Memory and Language, 34, 89-109.
³ They were actually really, really old idioms that were completely unfamilar to the participants.
⁴ Keysar, B., Shen, Y., Glucksberg, ?S, & Horton, W. (2000). Conventional Language: How Metaphorical Is It? Journal of Memory and Language, 43, 576–593.
⁵ McGlone, M. S. (1996). Conceptual metaphors and figurative language interpretation: Food for thought? Journal of Memory and Language, 35, 544–565.

The Internet and the Extended Mind

While we're thinking about the internet as distributed cognition, we might also think about it from the perspective of the active externalism. From this perspective, when entities external to the mind are used in the course of thinking, those entities are part of an extended cognitive system. Thus, in these cases, cognition takes place both inside and outside of the brain. As Clark and Chalmers put it:

Epistemic action, we suggest, demands spread of epistemic credit. If, as we confront some task, a part of the world functions as a process which, were it done in the head, we would have no hesitation in recognizing as part of the cognitive process, then that part of the world is (so we claim) part of the cognitive process. Cognitive processes ain't (all) in the head!

They go on to say:

In these cases, the human organism is linked with an external entity in a two-way interaction, creating a coupled system that can be seen as a cognitive system in its own right. All the components in the system play an active causal role, and they jointly govern behavior in the same sort of way that cognition usually does. If we remove the external component the system's behavioral competence will drop, just as it would if we removed part of its brain. Our thesis is that this sort of coupled process counts equally well as a cognitive process, whether or not it is wholly in the head.

For the most part, the focus of Clark and Chalmers is on immediate interactions with the environment, as when we physically rotate objects rather than mentally doing so, or when we count on our fingers (using them, in their view, as a form of external working memory). However, I think one of the more interesting aspects of the internet is the way in which it serves as a sort of external long-term memory store. I think we can view the use of the internet for the storage and retrieval of information as a form of active externalism, or an instance of the extended mind, as well.

The primary criterion Clark and Chalmers discuss for treating the use of the external environment during cognitive tasks as part of an extended cognitive system requires that the external entities serve the same purpose as internal processes (i.e., the external and internal processes are causally systematic). I think the use of the internet as a long-term memory store meets this. Take search engines, for example. Often we may read something on the internet, and encode the gist of it. However, the details are either not encoded, or weakly encoded. We don't need to encode them. Using our knowledge of the gist, we can simply search the internet using keywords (retrieval cues), and quickly retrieve these details. We can then use this information to structure our knowledge of the domain on-line. We may even use the internet to store and retrieve information about ourselves. Blogs, for instance, serve as extended memory stores for our previous ideas, and articles we've read. Searching their archives, or using search engines to find previous posts, can therefore be an economic way to retrieve information about ourselves.

In these situations, the external information serves the same purposes that internal information in memory would. Much as we would when retrieving internal memories, we use retrieval cues to search for information, and that information then serves to structure our representations. Furthermore, the retrieved external information may serve to cue the retrieval of further autobiographical memories about inferences we had previously made based on that information, much as the retrieval of stored representations would. Thus, the internet, as a long-term memory store, appears to be part of an extended cognitive system used for retrieving and reasoning about information in extended long-term memory.

The Internet as Distributed Cognition

The word "meme" is one of the most pervasive memes in the blogosphere, and rightly so. Ideas, phrases, talking-points, etc., often catch on, or don't, and gain a life of their own in blog posts and internet articles. One of the more recent examples is the "moral values" issue which has somehow come to be viewed as the issue of the election, despite the fact that there is no empirical evidence to indicate that this issue was any more influential in 2004 than it has been in the past. While it may be productive to use the meme metaphor when thinking about the ways in which such ideas become distributed throughout the minds of many individuals, I think there's another way of thinking about it that might capture more of what's going on. In many ways, the development of such ideas (and not merely their proliferation) is similar to the sorts of phenomena discussed within the distributed cognition approach to cognitive phenomena. This approach doesn't just highlight the ways in which ideas are transmitted and reproduced, but also explains the ways in representations of and reasoning about them develops over time and across individuals. It might be interesting, then, to treat the blogosphere, and the internet in general, as distributed cognition.

Distributed cognition is a fairly new idea (first proposed in the late 80s, but only recently gaining some popularity) that is opposed to the traditional Cartesian views of the mind that are prominent even among materialists in philosophy of mind. Here is a short description of the D.C. approach from Yvonne Rogers and Mike Scaife:

The distributed cognition approach is concerned with cognitive phenomena that cover a wide spectrum; from analysing the properties and processes of a system of actors interacting with each other and an array of technological artefacts to perform some activity (e.g. flying a plane) to analysing the properties and processes of a brain activity (e.g. perceiving depth). To date, however, most attention has focused on cognitive systems of work practices, like cockpits, air traffic control (Halverson), software teams (Flor) and engineering (Rogers).

The Distributed Cognition approach emphasises the distributed nature of cognitive phenomena across individuals, artefacts and internal and external representations in terms of a common language of ‘representational states’ and ‘media’. In doing this it dissolves the traditional divisions between the inside/outside boundary of the individual and the culture/cognition distinction that anthropologists and cognitive psychologists have historically created. Instead, it focuses on the interactions between the distributed structures of the phenomenon that is under scrutiny.

For the most part, researchers studying distributed cognition focus on work tasks, such as flying a plane or navigating a ship. However, the same types of principles may also be applied to other types of reasoning and behavior as well. For instance, in the blogosophere, the ways in which individuals represent and reason about certain concepts is often a product of the dynamic interactions of multiple individuals over time. This dynamic exchange of ideas can create products (new concepts, or new representations of old concepts) that had not existed prior to the interaction of multiple minds. In essence, blogospheric discussions can be viewed as attempts to solve problems, and the problem-solving behavior is distributed across multiple minds and external media.

You might be thinking that this is just the way all interactive dialogue (and multilogue) works to produce epistemic and cultural innovations, but that is really my point. Instead of thinking about dialogue in the traditional ways, in which individual, informationally-encapsulated minds interact through external symbols, we can treat these instances of interactive reasoning as distributed cognition in which multiple minds are intertwined across time. Also, instead of thinking about these interactions from the perspective of the selfish reproduction of the concepts and symbols themselves, as the memetic approach does, we can focus on the ways in which the distributed aspects of the representing and reasoning about these concepts and symbols reproduces and recreates them, reaching novel solutions to difficult conceptual problems.

UPDATE: Brandon from Siris alerted me to this post of his on the same topic. It's better than mine, so you should go read it instead.

Homo floresiensis is E.T.

No, H. floresiensis isn't from outter space. Instead, this new human species isn't really a new species at all, but a tiny Homo Sapien like the woman who played E.T. in the Spielberg film. At least, that's what WorldNetDaily tells us. Apparently the article's author, Kelly Hollowell, thinks that "modern-day dwarfs" have brains that are "closer to the brain size of a modern day monkey and other pre-human ancestors who purportedly became extinct 2 million years ago," too. At least, that's what I get from the comparison. Do you ever wonder where WND and Tech Central Station find these people?

Accidental Surfing

I've recently begun using Mozilla Firefox instead of Netscape or Internet Explorer. I'm used to just typing in a few letters of a url that I have visited, and then clicking on the full url in the history window. However, sometimes I miss, and Firefox has this interesting feature in which it apparently tries to guess which url I meant based on the letters. This takes me to some interesting sites. I've accidentally run into all kinds of businesses, personal sites, and porn. Today I accidentally ended up at the weirdest site yet. I'm not even going to describe it. If you're curious, you're just going to have to go. Here it is.

Metaphor IV: The Reckoning

In the final installment of Mixing Memory's metaphor series (for now -- at some later date I'll get to novel vs. dead metaphors), I try to use the empirical data to distinguish between the categorization and structure mapping theories of metaphor. Before I start, I should make it clear that there is certainly not a consensus among researchers about which model is the correct one, though my feeling is that most are in the comparison camp, rather than the categorization camp, even if they don't fully buy the structure mapping account. Part of the problem is that it is difficult to distinguish between the two accounts, and they are both powerful enough to handle most of the data out there.

Here's an example: In a 1987 paper¹, Kelly and Keil asked participants to rate how semantically different concepts from two disparate conceptual domains were. After that, they presented the participants with metaphors containing those concepts. After rating the aptness of the metaphors, participants were against asked to rate the semantic difference between the concepts. Concepts that had been in highly rated metaphors were then rated as more semantically similar than they had been in the first ratings. People in the comparison camp take this as evidence that the two domains were mapped during metaphor comprehension, and thus are more similar after the mapping. However, one could easily interpret this result as evidence for the categorization view. Since, under this view, the vehicle and topic are placed in the same category, and since intra-category similarity is usually higher than inter-category similarity, it stands to reason that similarity ratings would be higher after the categorization than before.

Despite this difficulty in distinguishing the two types of theories empirically, I think they can be distinguished. Furthermore, I think we can decide between the two right here and now. As I said in the post on the attributive categorization theory, in its most recent form the categorization view of metapor involves a comparison, with the unintuitive categorization process on top. The purpose of this process seems to be to explain the irreversability of metaphorical statements. Since the categorization theory now involves comparison, and since the categorization aspect itself is unintuitive, if we can come up with a comparison theory of metaphor that can explain this irreversability, then we can do away with the categorization theory altogether. The question we have to ask, then, is can structure mapping, the most prominent comparison theory, account for the irreversability?

The answer to that question is a little bit more complex than it seems. To answer it, we first have to determine when the irreversability arises. If it arises from the very beginning of the comprehension process, then only the categorization theory can account for it. However, if it arises over the course of the comprehension process, then structure mapping can account for it. This is because in structure mapping, once the mapping has been made, the inferences can only move in one direction. However, when the mappings first start, the two domains are treated equally, and thus there exists a symmetry between the two roles in the metaphor, so that the topic and vehicle could be reversed without a problem.

To test when metaphorical statements become irreversable, Wolff and Gentner² used the "true/false" task described in the first post. To recap, in this task, participants are asked to determine whether sentences are literally true or false. A third of the sentences are literally true, a third are literally false and do not make sense as metaphors (sans context), and a third are metaphors that received high scores on an aptness scale. Participants quickly rate the literally true statements as true, and the abhorent literally false and aberrant statements as false. However, they took significantly longer to rate the metaphorical statements, which are also literally false, as false. This is generally taken as evidence that metaphors are processed automatically, and without prior literal processing. In this case, the experiment is used because it taps into the early processing of metaphors, presumably pre-mapping. If it can be shown that participants have the same trouble rating metaphors as literally false even when the topic and vehicle are reversed, then this would be strong evidence that metaphors are not irreversable in the initial processing stages. This would in turn be evidence that the structure mapping theory can account for the irreversability of metaphors, thus rendering superflous the categorization phase in the attributive categorization theory. In two experiments, this is in fact what Wolff and Gentner found.

So now that we've reached the end of our short journey through cognitive scientific views of metaphor (glaringly ommitting cognitive linguistic theories), I think we can come to a pretty firm conclusion: metaphors involve comparisons. Once we've shown that a comparison theory can account for the irreversability of metaphor, we no longer need to posit a categorization phase in metaphor comprehension. Is structure mapping the right theory to model the comparison process in metaphors? That's a more difficult question to answer. Right now, the only other viable (and I use that word very, very loosely) theories are in cognitive linguistics. In fact, blending may be the only other viable theory currently available, and it is not inconsistent with structure mapping theory (you could use structure mapping theory to acheive everything that blending theories say is going on). So, as things look, structure mapping theory is the best cognitive theory of metaphor we've got. As you might expect from an algorithmic theory in a young discipline, there are plenty of problems with it. Still, it has survived 20 years of empirical tests in the areas of analogy and similarity, and until we have something better, it will probably continue to produce the best experiments on metaphor comprehension.

In case you missed them, Metaphor posts I-III can be found here, here, and here.

¹ Kelly, M., & Keil, F . C. (1987). Conceptual domains and the comprehension of metaphor. Metaphor and Symbolic Activity, 2, 33-51 .
² Woff, P. & Gentner, D. (2000). Evidence for Role-Neutral Initial Processing of Metaphors. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26(2), 529-541.

It's the End of the World as We Know It

The liberal corner of the blogosphere is filled with "What happened," "What now," and "Oh fuck!" posts. Not wanting to be left out, I feel I have to offer my own post-election post. First, a few highlights from the others.

Kevin Drum:

The lefties will say we need to stop trying to be Republican Lites, the DNCers will say we need to move to the center, the New Republic will say we need to get serious about national security, Amy Sullivan will say we need to pay more attention to religion, George Lakoff will say we need better issue framing, the Washington Monthly editors will say we need a more potent vision, etc. etc. I'm not sure who's right, but we'll figure it out.

But one thing not to do is hide under the blankets and give up. We lost an election, that's all. There will be another one in a few years, and if we persuade a few more people that we're right, we'll win it. Tomorrow would be a good day to start doing that persuading.

John Quiggin:

While I’ve tried to be open to more optimistic possibilities, it’s far more likely that the second Bush Administration will be more of the same, and worse. The problem for the winners is that the consequences of the Administration’s policies, still debatable in 2004, will be grimly evident by 2008, and there will be no one but Republicans to take the blame. In purely partisan terms, as I argued several times before the election, this was a good one to lose.

Scott Lemieux:

The next four years, to state the obvious, will be ugly. Nothing good comes from this. I don't, for a second, relish the thought of Bush being left to clean up his mess. A lot of people will suffer, and path dependencies that will make even modest progressive reform difficult will be further entrenched. It will be a very, very, difficult struggle.

PZ Myers:

I fear for the future. The Republican party has established a solid base in America’s strengths: fear, ignorance, and swagger. The Democrats failed to win by opposing those ugly values; will they, too, resort to pandering to them in the next election? Will the lesson they learn be that progressive ideals must be sacrificed to make political gains?
I worry about my kids, and the children of those folks in Red State America who think safety lies in blithely handing a blank check to ideologues. How much of their blood will have to be spilled in self-destructive wars? How great a burden of debt will they have to bear, in order to guarantee that today’s wealthy are sufficiently comfortable? When the Supreme Court is loaded with mullahs of the religious right, what liberties will be lost to them?

The Anonymous Lecturer:

Here's my prediction for the next four years: the cultural conservatives will seek to Christianize the US completely. Anyone who dissents, criticizes or even asks questions about the neo-conservative/Christian right controlled police society will be labeled a terrorist or terrorist sympathizer. Welcome to the new "red scare" people. We're in for a ride.

Brian Leiter:

The results of the U.S. election were a resounding victory for fascist theocracy and war-mongering. A philosopher from England, who wrote to me this morning, no doubt expresses the view of the civilized world about Bush's victory: "The brains of a hamster, the religious and moral views of a savage, the record of an almost complete failure. And yet the winner of the popular vote."

With the exception of Drum's comments, the theme of most of the posts I've read today has been that Bush's reelection is a disaster of epic proportions. John Quiggin thinks that this disaster will ultimately benefit Democrats politically, but it's still a disaster. Leiter goes so far as to call it "a resounding victory for fascist theocracy and war-mongering," which expresses in much more histrionic form the other part of the view I've seen, both from the Left and from the Right. It's expressed best by William Bennett, when he says:

Having restored decency to the White House, President Bush now has a mandate to affect policy that will promote a more decent society, through both politics and law. His supporters want that, and have given him a mandate in their popular and electoral votes to see to it. Now is the time to begin our long, national cultural renewal (“The Great Relearning,” as novelist Tom Wolfe calls it) — no less in legislation than in federal court appointments. It is, after all, the main reason George W. Bush was reelected. (via Crooked Timber)

I am not at all surprised that conservatives like Bennett see it as a victory for their regressive religious and social agenda. That's all they can see it as. However, I am very surprised to see so many liberals thinking that conservative hillbilly morality had anything to do with Bush getting elected. Sure, about half of the people who vote Republican no matter what (you know, the 40% of the population who would vote Republican if Marx had an R next to his name on the ballot, and then be able to create a fairly good story to convince themselves that they did so for principled reasons) are fundamentalist evangelical Christians. The other half is a mish-mash of fiscal conservatives, social conservatives, foriegn policy conservatives, and the like. Still, it's not this 40% of the population, composed of these two halves, that got Bush elected. He was elected because more than half of the 20% of the people who don't vote R or D every time voted for him (I'm assuming, and I think this is the case, that the turnout wasn't really more favorable for Republicans). Why did they do this? Some might have done so for social reasons, but I would bet a lot of money (if I had it) that the vast majority of the members of that group who voted for Bush did so because we are at "war," and Bush's election therefore says very little about the moral mood of the country that we didn't already know.

Here's a more detailed description of what I think happened with those voters. Some of them voted for Bush because they felt the evil we know is better than the potential evil we don't know. Some voted for Bush because they bought the absurd premise that even though Bush has bungled the aftermath of two wars, he can do a better job in the future than Kerry could (Kerry might have done just as poor a job, but he couldn't have done much worse). Some voted for Bush because they failed to understand that Kerry's voting for the war, and later opposing it, might have been the result of looking at the facts as they became available, rather than political posturing or an inherent indecisiveness on Kerry's part. Some voted for Bush because they are so scared of terrorism that they refused to recognize Bush's lies and mistakes, and that Bush has made the world more dangerous, and voted from fear rather than reason. I could go on, but you get the point. The bottom line is, the people who gave Bush a majority, and four more years in office, did so because of the war, not because they agree with his social agenda (which is not to say they disagree with it).

Is the apocolypse at hand, as so many liberals seem to think? I doubt it. Certainly, the next four years will be a mess. The economy will likely suffer, at least for those of us who don't make enough to buy a new Mercedes every year; things will likely not get better in Afghanistan, Iraq, or the "War on Terrorism;" our relations with other countries, especially our traditional allies, will further deteriorate, science will remain under constant assault, and a fundamentalist Christian social agenda will be further advanced. But this doesn't mean the end of the world. It just means that we liberals have our work cut out for us. I wish I could say I had confidence in our ability to accomplish what we need to accomplish, but I don't. The Democratic Party is simply not a vehicle capable of getting us to our desired destination. Furthermore, as this election shows, no matter how badly things get screwed up for four years under Republican leadership in all three branches of government, the American people are not capable of placing any blame on them. So is this the end of the world? No, but the world just became a lot more fucked up. Now back to metaphors.

UPDATE: You know how I said that only about half of the people who vote Republican in every election, no matter whose running for what, are really doing so for "moral" reasons? Since about 40% of voters fit this description, that would mean 20% of the voters are voting Republican time in and time out for moral reasons. Well, I was right. See the exit poll here, in which 22% listed "moral values" as the issue most important to them. (via Majikthise) Focusing on this people is counterproductive for liberals. They will never, never ever ever ever, vote for anything resembling a liberal agenda.

Metaphor III: Metaphor Is Categorization

I have heard that there is an election today, and I've heard that it's going to be close and contentious, but I don't care. Here at Mixing Memory, we're only worried about metaphor for now (and soon, classical vs. connectionist architectures, and perhaps after that, idioms, and after that... the sky's the limit). In the first two metaphor posts, I talked about the history of cognitive theories of metaphor, and the structure mapping theory of metaphor. Now it's on to the other prominent view of metaphor, one that differs almost entirely in its description of what metaphors are. Throughout most of the history of western philosophy, and certainly throughout the history of cognitive science, metaphors have been viewed as comparisons. This was the view that Aristotle took, it's the view of C. S. Peirce (as far as I can tell, and thanks to Clark for the tip), and it is the view of Black, Ortony, the cognitive linguists (shudder), and the structure mapping theoriests. However, there have been criticisms of this account among some cognitive psychologists. Most notably, Glucksberg and his colleagues have claimed that the comparison view of metaphor does not fully capture the assymetry inherent in all metaphorical statements. Most comparisons are, in fact, asymetrical. The classic example of asymetry in comparisons comes from work by Amos Tversky. He showed that participants judged "North Korea is like Red China" to be better comparisons than "Red China is like North Korea," because the features that are relevant to the comparison (e.g., their status as communist states) are more salient in one domain (China) than in the other (Korea). Furthermore, Korea shares a higher portion of its features with China than China does with Korea. However, the assymetry of such comparisons is not absolute. One can say that China is like Korea, and get away with it. However, unlike ordinary comparisons, metaphors are not reversible. While "My lawyer is a shark" makes sense, "The shark is a lawyer" does not. This irreversability is not captured by comparison theories of metaphor, according to Glucksberg, and the failure to do so demands a new approach to metaphor. So now, instead of viewing metaphor as comparison, metaphor is treated as categorization. Taxonomic category relationships are irreversable. A robin is a bird, but birds are not robins. Thus, treating metaphors as categorization statements captures the extent of the asymmetry in metaphor.

The gist of the categorization theory of metaphor is fairly simple. Metaphors are is-a, or class-inclusion statements in which the topic is said to be a member of a category represented by the vehicle. The vehicle itself is not the category into which the topic is placed. Instead, the vehicle is chosen because it is a member of the category that exemplifies the category's defining features. In some cases, the category is an existing category, but in most cases, it will have to be produced on-line during the processing of the metaphor. For example, in the metaphor "My job is a jail," the vehicle, jail, is chosen because it is a salient member of a category created specifically for the metaphor. In this case, the category created specifically for the metaphor would be something like "confining places." Because the vehicle stands for both the concept to which it ordinarily refers, and the category into which the topic is being placed, the vehicle is said to have "dual reference." The topic, in turn, as a member of the category, inherets all of the attributes of that category.¹

The most recent version of the categorization view of metaphor, the attributive categorization theory², the vehicle and topic interact to select which properties of the vehicle will be used to select the category the vehicle represents, and into which the topic will be placed. Once again, then, the attributive categorization theory resembles Black's interactive theory of metaphor. This interactive aspect of the categorization process helps solve two problems with the earlier categorization theories. The first is the problem of how we know which of the vehicle's features references the category. In this case, the topic itself helps to select these features. The second problem is the fact that any given vehicle can be used to represent different categories in different metaphors. By allowing the topic to select which features of the vehicle are relevant, the vehicle is then free to represent different categories when it is paired with different topics.

It's interesting, in the end, that even when metaphor is treated as categorization, a comparison is still needed to select which features define the category into which the topic is placed. As in all of the previous theories of metaphor, the topic must be compared to the vehicle in order to determine the relevant features in the vehicle. The categorization aspect of the theory mainly serves the purpose of creating the irreversability of the metaphor. Otherwise, the categorization process would be superfluous, and if a comparison theory of metaphor could create the level of asymmetry required of a theory of metaphor, this would probably render categorization theories obsolete. Introspectively, there doesn't seem to be any sort of categorization going on when we process metaphors. While this is not, of course, a damning charge for any cognitive theory, because most of the work in most cognitive tasks is going on unconsciously, it does make the categorization view feel a little odd. For these reasons, the categorization theory has not been widely accepted by theorists who have held comparison views of metaphor all along. A local war has erupted between the structure mapping and attributive categorization camps, and there's a lot of empirical evidence out there. In the next post, I'll try to cover some of it.

¹ Glucksberg, S ., & Keysar, B . (1990) . Understanding metaphorical comparisons : Beyond similarity. Psychological Review, 97, 3-18.
² Glucksberg, S ., McGlone, M. S., & Manfredi, D . (1997) . Property attribution in metaphor comprehension . Journal of Memory and Language, 36, 50-67 .

Metaphor II: "Metaphor Is Like Analogy"

Onward we go to the first contemporary view of metaphor, structure mapping theory. Before I start, though, I want to clear something up. Perhaps no one has actually been confused, but I'm afraid that I haven't made something clear that should be made clear. For the most part, cognitive theories of metaphor -- cognitive linguistic accounts, which purport to be theories of all cognition and, if the cognitive linguists had their way, would also combine to serve as the Unified Field Theory, aside -- are intended to account for common, everyday uses of metaphor. "My surgeon is a butcher" and "my lawyer is a shark" are great examples of this sort of metaphor. "A poem should be... silent as the sleeve-worn stone of casement ledges where the moss has grown" is not a good example of this sort of metaphor. Cognitive scientists usually call this latter sort of metaphor "creative metaphor," and to be honest, these types of metaphor have been neglected in cognitive theories. Unlike commonplace metaphors, creative metaphors are often incompatible with comparison and categorization views of metaphor, and more importantly, they are not easily or automatically processed. In fact, I suspect that most of them are specifically designed to induce deliberative thinking, at the same time they elicit automatic emotional and imaginative responses. So, when I talk about metaphor, I mean commonplace, not creative metaphor. I wish we had a viable cognitive account of metaphor that could capture "an empty doorway and a maple leaf" and "hills like white elephants," but we don't.

OK, now that that's out of the way, I can get started. The structure mapping theory of metaphor treats metaphors as analogies, at least in their underlying cognitive mechanisms. Some metaphors are obviously similar to analogies, and may even be considered analogies. "Encyclopedias are gold mines" (a common metaphor in the cognitive literature), for instance, clearly involves the mapping of relational structure between the encyclopedia and gold mine domains. Other metaphors are less obviously analogical. "My lawyer is a shark" seems primarily designed to map a few specific attributes of sharks onto my lawyer, in order to highlight those attributes in my lawyer. It is thus more like a literal similarity comparison (e.g., "Alligator meat is like chicken") than analogical comparisons (e.g., "The atom is like the solar system."). On the surface, the existence of these two different types of metaphor seems to make the possibility of a general theory of metaphor that treats metaphor as analogy (at least in terms of processes) impossible. However, it turns out that literal similarity comparisons may also involve the same processes as analogies, which means that metaphors that are like literal similarity comparisons could also be like analogies.

Here is the gist of the theory: metaphor is like analogy. Analogies involve the "structural alignment" of two (or more) structured representations (representations containing objects, their relations, and their attributes, along with relations between relations) so that the common elements in the representations are mapped onto each other¹. Structural alignment occurs under three primary constraints: systematicity, one-to-one mapping, and parallel connectivity. Systematicity requires that, all things being equal, higher-order mappings are preferred. This means that mappings involving relations between relations will be preferred to mappings involving relations between objects, and mappings between relations between objects will be preferred to mappings between objects or their attributes. The one-to-one mapping constraint requires that each element in a representation be connected to at most one element in the other domain. For instance, in "The atom is like the solar system" analogy, once we map the planets in the solar system domain onto electrons in the atom domain, we cannot also map the planets onto the nucleus or some other element in the atom domain. The third constraint, parallel-connectivity, requires that when elements are mapped onto each other, their arguments are also mapped. For instance, when we map the "Revolve around" relation in the "Atom is like the solar system" analogy, then parallel connectivity requires that the arguments (planets-sun in the solar system domain, and electrons-nucleus in the atom domain) be mapped as well. These constraints allow analogical comparisons to preserve the maximum amount of common structure between the two (or more) domains being compared, and this in turn makes for easier and more productive inferences, which are what motivates most analogies in the first place.

While structure mapping theory was originally intended as a theory of analogy, it can also be extended to literal similarity comparisons like "Alligator meat is like chicken," which are designed to highlight common objects or attributes, and not common relational structure. To do this, the mappings are restricted to objects or attributes². Since metaphors resemble both types of comparisons, structure mapping has, over the last decade or so, been used as a theory of metaphor. To see how this works, I'm going to let the theorists themselves describe an actual example, because I know I couldn't do it any better. Here is a passage from Bowdle and Gentner (In Press)³:

To better illustrate this approach to metaphoric mappings, consider Socrates was a midwife - a metaphor that was first used in Plato's Theaetetus,, and that has been examined in depth by Kittay and Lehrer (1981)... Structure-mapping theory and SME [Structure Mapping Engine, the computational implementation of Structure-mapping theory] predict the following sequence of events during the interpretation of the metaphor. First, the identical predicates in the target and base concepts (i .e., the relations helps and produce) are matched, and the arguments of these predicates are placed in correspondence by parallel connectivity : midwife --> Socrates, mother --> student and child --> idea. Next, these local matches are coalesced into a global system of matches that is maximally consistent . Finally, predicates that are unique to the base but connected to the aligned structure (i.e., those predicates specifying the gradual development of the child within the mother) are carried over to the target . Thus, the metaphor could be interpreted as meaning something like, "Socrates did not simply teach his students new ideas, but rather helped them realize ideas that had been developing within them all along". (p. 11).

There you have it: the two domains (Socrates and midwife) are aligned so that their common relational structure (Socrates helps the student produce an idea; the midwife helps the mother produce a child) is in correspondence. After the mapping occurs, information from the vehicle is carried over to the topic in the form of inferences, so that we now see Socrates as helping give birth to ideas that had been developing in the minds of students, as the midwife helps give birth to children that had been developing inside of mothers.

I won't get into the evidence for this view until I've posted on the other major theory of metaphor, but since I described the assymetry of metaphorical statements as being one of the most important features for a theory of metaphor to capture, in the first post, I will quickly describe how structure mapping explains this assymetry. Like metaphors, analogies are always assymetrical. The primary purpose of analogy, in most cases, is to compare a lesser-known domain (e.g., the atom) with a better-known one (e.g., the solar system). This allows one to carry structure from the better-known domain over to the lesser-known domain, in the form of inferences, to produce more knowledge about it. This sort of directional production of inferences is what produces the assymetry in metaphors as well. In metaphor, the vehicle corresponds to the better-known domain, and the topic to the lesser-known, and inferences are produced from the vehicle to the topic. The "Socrates was a midwife" metaphor demonstrates this. The inferences about internal development are carried from the vehicle to the topic, and no inferences are made in the other direction.

Finally, as I said in the first post, this theory of metaphor was inspired, in part, but Black's interactive theory. The similarity of this theory of metaphor to Black's interactive theory comes from the fact that it is the interaction between the two domains, the vehicle and topic, in the form of the alignment of common relational structure, that produces the relevant features to carry over from the vehicle to the topic. Unlike Black's theory, however, the way in which the interaction determines the relevant features is made explicit. In fact, the mechanisms for discovering these features are explicit enough for the computational implementation of structure mapping theory, the Structure Mapping Engine (SME), to discover them on its own, and thus produce interpretations of metaphors similar to those of human subjects, without human intervention after the encoding of the initial representations of the two domains.

In the next post, I will detail the attributive categorization model, and after that, we'll get to real empirical evidence. In thinking about what might come after that, I've decided that a discussion of novel vs. dead or conventional metaphors might be interesting, because it has implications for the two major theories. Perhaps after the election I'll talk about cognitive linguistic theories of metaphor, too.

¹ Gentner, D., Bowdle, B., Wolff, P., & Boronat, C. (2001). Metaphor is like analogy. In Centner, D., Holyoak, K.J., & Kokinov, B.N. (Eds.), The analogical mind: Perspectives from
cognitive science (pp. 199-253). Cambridge MA, MIT Press.
² In most metaphors, even those that are ostensibly about specific attributes (e.g., "My lawyer is a shark" is about "aggression," or some similar attribute), there is also relational information that can and will be mapped in the process of understanding the metaphor. However, structure mapping theory can handle similarity comparisons and metaphors that only involve the mapping of attributes.
³ Bowdle, B., & Gentner, D. (in press). The career of metaphor. Psychological Review.