I’ve been wondering this morning to what extent, if any, an ability to predict events is a crucial measure of intelligence? Maybe even the crucial measure?
For instance, can we measure how “people smart” someone is by how likely they are to accurately predict the behavior of others? Or, how “self smart” someone is by how likely they are to accurately predict their own behavior?
Again, can we measure how “logic smart” someone is by how likely they are to accurately predict the next logical step in a sequence? Or, how “body smart” someone is by how likely they are to accurately predict their or someone else’s next motion in a sequence of movements?
On the face of it, the notion of predictive prowess as a crucial measure of intelligence seems to fit nicely with a certain bit of epistemology. We know, for instance, that our senses do not reproduce reality but instead represent or symbolize it. That is, we do not see photons bouncing off atoms at a certain wavelength (which is the reality) — instead we see a certain color (which is merely a representation or symbol, created by our brain, of the reality).
Now when we move up a level or two — from the senses to conscious thought — we once again find symbolism, for conscious thought does not reproduce or mirror reality, but instead represents or symbolizes it. A point I previously discussed here. Indeed, it seems as if on every level we know of, the brain creates symbols that have the same basic relationship to reality that the symbols on a map have to the terrain that the map represents or symbolizes.
Now, the notion of predictive prowess as a crucial measure of intelligence seems to fit nicely with all that epistemology I just referred to. But how so?
Well, the way you judge whether a map is accurate is essentially the same way you judge whether a proposition is true. Namely by whether the map predicts its terrain or by whether the proposition predicts what it refers to. Suppose a map says an erotic dance club can be found at location X. We travel to location X and do indeed find an erotic dance club there. Thus, in a state of sublime happiness, we pronounce the map “accurate”. In much the same way, the truth of a proposition is measured by the extent to which it predicts what it refers to.
Suppose I assert the proposition that Jones will pass by the wallet laying on the sidewalk without investigating. Suppose Jones does indeed pass by the wallet without pausing to investigate it. My proposition is demonstrated true because it turned out to be an accurate prediction of what Jones would do.
Again, suppose I assert the proposition that I am not a malicious person. Suppose then I go from site to site on the internet trolling people with threats and insults. My proposition is demonstrated false because it turned out to be an inaccurate prediction of my own behavior.
Can it therefore be reasonably said that a crucial measure of intelligence is how well a person (or species, for that matter) is able to predict events?
Well, it’s a bit difficult to argue the converse, isn’t it? It seems to make a mockery of what we might mean by “intelligence” to argue that an inability to predict events is a mark of high intelligence. As we’ve seen, that would be much the same as saying that “intelligence” consisted of an inability to formulate true propositions.
Of course, the absurdity of the converse does not necessarily mean our thesis is true, because we could define “intelligence” in ways other than an ability to accurately predict events (or the converse of that). Still, I find it intriguing to consider that intelligence — perhaps even the best possible definition of intelligence — might be described as the likelihood of someone’s making an accurate prediction of events.
The other notion I’ve been toying with this morning consists of the claim that “intelligence” can be elegantly defined as the total amount of caffeine one has coursing through one’s brain. At the moment, I’m about to get my second cup of coffee and further ponder that notion.
















My StumbleUpon Page
10 responses so far ↓
Lirone // May 27, 2008 at 8:46 am
Interesting idea … this is certainly part of intelligence, though I don’t feel that it’s the whole story. A computer model can predict the future, but I wouldn’t necessarily call that intelligent.
I think one of the crucial parts of intelligence is the ability to make links and detect patterns - which is vital for experiencing the present as well as predicting the future .
There’s a fascinating series about the biases that causes us to make wrong predictions, particularly about our own future wants and desires, and how to correct for this, over at PsyBlog (a site I thoroughly recommend)
meleah rebeccah // May 27, 2008 at 11:41 am
I take a lot of pride in being:
“self smart” -by how likely they are to accurately predict their own behavior?
I spend quite a bit of time self analyzing.
Paul // May 27, 2008 at 4:18 pm
Meleah, I’m proud to say I formed that opinion of you some time ago when reading “Off the Pole”. I was impressed with how honest, detailed, and insightful was your self-knowledge. It was like being in your shoes.
Paul // May 27, 2008 at 4:21 pm
Thanks for the link to Psyblog, Lirone!
“I think one of the crucial parts of intelligence is the ability to make links and detect patterns - which is vital for experiencing the present as well as predicting the future .”
I’m very curious about what you said there. Would you elaborate please on how making links and detecting patterns is vital for experiencing the present?
Dana Hunter // May 28, 2008 at 6:25 am
I love the idea that amount of caffeine in brain = intelligence. If that’s true, it’s official: we’re ruddy geniuses! LOL.
Bring me another Coke and we’ll go take over Mensa!
lirone // May 28, 2008 at 12:23 pm
My answer turned into a post of its own…
Patterns, intelligence and the present
Paul // May 28, 2008 at 4:07 pm
@ Dana:
@ Lirone: Thank you for such a thoughtful response! I’ve been mulling over what you’ve said and will respond on your blog.
Erik // May 30, 2008 at 10:32 am
Paul/lirone,
I think you are both right, but you are talking about different parts of the whole. Paul addresses predictions as a measure of intelligence, while lirone is addressing the underlying mechanisms whose function leads to the output that we label “intelligence.” Paul alludes to this when he mentions predicting the next logical step in a sequence. To make an accurate prediction one must determine the underlying pattern (generally done by noting links between individual elements of a pattern). Thus someone judged to be of high intelligence is likely to accurately integrate a larger number of elements to produce a better prediction more rapidly than someone judged to be of lower intelligence.
Ability to recall information is useful in generating predictions because it allows one to hold more disparate elements and/or a longer historical string of elements in mind while recognizing the pattern, deepening the granularity of the pattern and improving the pattern’s predictive power. Thus the depth and ease of recall of knowledge is also often considered a measure of intelligence. I’d argue that accuracy of prediction is a more useful measure, however; we’ve all heard of people who are “book smart but practical dumb.” While these people may have huge knowledge bases, they aren’t able to make effective use of the knowledge by creating predictions that can be used to make high quality decisions, so they are disparaged.
Practice generating predictions can also lead to improved speed and quality of future predictions of similar type, perhaps explaining why people with less education, or people with less training in critical thinking, are often believed to be less intelligent.
This idea doesn’t seem new, although I admit it’s the first time I’ve really thought about it. I never read any of the “develop your emotional intelligence” books or their ilk since they seemed too pop-psych for me, but I’d bet that in their hundreds of pages they basically say what Paul and lirone have outlined here so cogently.
Which leads me to my sole disagreement with lirone: based on these criteria and building on Paul’s notion of different kinds of “smart” (logic smart, body smart, etc.) a computer program that generates a model clearly can be considered intelligent, at least with regard to the area of prediction addressed by its output. In areas outside its output it would be a moron. If I’m right then it is reasonable to consider a modern chess program more “intelligent”, at least with regards to chess, than early programs (and me, too, incidentally) because it more accurately integrates the pattern of pieces to generate a faster, more useful prediction of the move that will most likely generate a win.
Dang. I always try to write short comments and so rarely succeed. Sorry.
lirone // May 30, 2008 at 10:48 am
Erik… I have the same problem on long comments, which is why I often resort to writing a separate post!
I understand where you’re coming from on the computer models, though for me the qualification you make “at least with regard to the area of prediction” is for me a very important limitation.
I would say that intelligence is a generalisable ability to identify patterns and make predictions rather than a specific one, so a programme that just operates one model doesn’t quite make it for me.
So I would say an ability to make predictions _in varying situations_ would for me count as a measure of intelligence - because it demonstrates that underlying pattern recognition ability.
Erik // May 30, 2008 at 11:30 am
lirone,
Yeah, I keep thinking I should start my own blog, but I’m just too lazy to try to come up with topics every day.
I completely agree that the more generalizable the ability to identify patterns and make predictions the more someone (something?) should be considered intelligent. Versatility is definitely important, so much so that I would consider the best predictive program to be, overall, a moron even compared with many non-human animals. I’d argue that the programs still have measurable intelligence, however. Even a crumby chess program is likely to make a much more “intelligent” move than I would, for example.
Including generalizability/versatility in the definition of intelligence additionally creates the problem of measuring, and perhaps assigning a lower bound, of generalizability. This seems unparsimonious and unnecessary to me. In a sense I’d give “intelligence points” for the quality of predictions in every area that something can make predictions. The vast number of different areas in which most humans can produce predictions, even if those predictions are pretty poor, still brings us out ahead of computers. At least for now….
Leave a Comment