Google's PageRank Works Like Our BrainsWe've joked in the past about how Google effectively acts as a a secondary or "backup" brain for many people. However, perhaps it wasn't so much of a joke. New research on how human memory and recall works suggests that the process is quite similar to Google's PageRank in determining what things are more important and should be recalled first. Basically, Google's PageRank looks at "popularity," not just in terms of how many links a site gets, but also in terms of how popular those links are. Thus, if you get linked from a more popular site, that's more valuable than getting linked by a bunch of non-popular sites. It turns out that the brain does something similar in linking concepts, judging not just the popularity, but the popularity of the concepts linked to the concepts. In fact, using Google's PageRank turned out to be a better predictor of how a brain would prioritize words than more commonly known methods.
This could be an interesting finding for the artificial intelligence community. After all, many in the AI community have been trying to figure out how to make computers act more like human brains for years, and various brute force methods haven't worked all that well. Obviously, the AI world has worked on various neural net research for quite some time, but it's nice to see at least some confirmation from the psychology side concerning a way to match up brains and algorithms. A couple years ago, we noted that intelligence was often correlated to people who knew what to forget rather than trying to remember everything. What that really shows is that good brains are better at prioritizing and ranking the importance of something -- and that's exactly what PageRank is intended to do. So, now, we just need Larry Page to get back from his honeymoon and get to work on BrainRank. Or would that be PageBrain? Of course, it's also worth noting that with the rise of search engine spamming, rumor has it that Google doesn't use PageRank that much any more. Perhaps that just means that our brains are vulnerable to concept spamming as well...
The Web article on which the TechDirt post is based is here:
... Thomas Griffiths of the University of California, Berkeley, and colleagues ranked the “importance” of over 5,000 words using the same basic Google formula, called PageRank. But instead of Internet links, the researchers tallied mental “links” between words as reflected in answers given in word-association games by people participating in previous studies.The investigators found that a word’s “PageRank” was a good predictor of how often it would show up when people were asked to think of words that start with A, with B, and so on.
When it came predicting these results, “PageRank” beat two other seemingly reasonable ranking systems: tallies of how often words show up in ordinary writing; and a simple count of direct “links” to a word that doesn’t consider how many words, in turn, link to those linking words.
In the PageRank formula, a page gains “importance” based on how many other pages link to it. But links from pages that are themselves “important,” confer more importance than those that aren’t. Thus, importance can be thought of as flowing through the Web’s link network toward the most highly “linked-in” sites.
One explanation for the new findings, wrote Griffiths and colleagues, could be that connections among brain cells work similarly to Web links. Cells that are targets of many connections might become more active than others, in the same way that highly linked-in websites are deemed more important.
“Our approach indicates how one can obtain novel models of human memory by studying the properties of successful information-retrieval systems, such as Internet search engines,” the group wrote in the study, published in this month’s issue of the research journal Psychological Science. The study also suggests brain science might help design better search engines and data-retrieval systems, they added. “These problems are actively being explored in computer science,” they wrote, but “one might be equally likely to find good solutions by studying the mind.”
The original article, cited as Thomas L. Griffiths, Mark Steyvers, Alana Firl (2007)
Google and the Mind: Predicting Fluency With PageRank
Psychological Science 18 (12), 1069–1076. , can be found here.
Here is the abstract:
ABSTRACT—Human memory and Internet search engines face a shared computational problem, needing to retrieve stored pieces of information in response to a query. We explored whether they employ similar solutions, testing whether we could predict human performance on a fluency task using PageRank, a component of the Google search engine. In this task, people were shown a letter of the alphabet and asked to name the first word beginning with that letter that came to mind. We show that PageRank, computed on a semantic network constructed from word-association data, outperformed word frequency and the number of words for which a word is named as an associate as a predictor of the words that people produced in this task. We identify two simple process models that could support this apparent correspondence between human memory and Internet search, and relate our results to previous rational models of memory.
There are so many things wrong with this study.
Suffice it to say, the paper is based on false premises and pretty much infers a correlation of results as a correspondence of process or mechanism. And the crux of the study is that the researchers asked humans to behave like search engines rather than asking search engines to behave like humans. So of course there were correlations. Still, correlations don't mean much.
Actually, to be fair, the paper merely speculates about the correspondence of processes.
I don't have time to take the study apart for readers today. But I will soon. Stay tuned to this space for more.
RSS Feed
Comments (3)
Wow. Siva, that's a bit of an unfair attack on me, don't you think? The post was merely a lighthearted post about the study that I thought our readers would enjoy. I never said PageRank worked like brains. I said the study "suggested" the process was similar, and noted how that could be useful in thinking about further research in the area.
As for your claim that the researchers asked people to act like a search engine, that is true. But the study (and again, my post) made it quite clear that this was focused on the recall function of the brain, so that seems perfectly reasonable. It's early research into an area, so like all early research it's only going to make loose connections that will then be useful for further research.
You are free to disagree, but to attack me personally just seems out of place and totally uncalled for. I'm quite surprised, honestly. I would have expected better.
Mike, I don't see the personal attack on you. And given the context of this research blog, I would say Siva's criticism is not even directed at these particular researchers, but at a general tendency to naturalize Google/Pagerank and our relationship to it.
The problem with the idea that Google (maybe) works like we do, is that it treats the tool as something neutral, as simply a reflection of us. We know that this is crucial to Google's image - "Google is your friend", as the forums remind us - but we also know this isn't true. To give just one reason, even something as seemingly straightforward as recall is affected by (and made possible by) a social environment, emotion, fatigue, etc.
More to the point, naturalizing Google as 'like us' seems inappropriate at a time when it is busy building and arranging the Web as we know it. Maybe the relevant question for researchers is not whether Google works like our mind, but what cognitive theories are built into technologies like PageRank. This would at least point us in the direction of Google's limitations ;)
i dont know the page ranks work like brain. But it is good for our site.
Thanks