Internet's memory effects quantified in computer study

By Jason Palmer

Computers and the internet are changing the nature of our memory, research in the journal Science suggests.

Psychology experiments showed that people presented with difficult questions began to think of computers.

When participants knew that facts would be available on a computer later, they had poor recall of answers but enhanced recall of where they were stored.

The researchers say the internet acts as a "transactive memory" that we depend upon to remember for us.

Lead author Betsy Sparrow of Columbia University said that transactive memory "is an idea that there are external memory sources - really storage places that exist in other people".

"There are people who are experts in certain things and we allow them to be, [to] make them responsible for certain kinds of information," she explained to BBC News.

Co-author of the paper Daniel Wegner, now at Harvard University, first proposed the transactive memory concept in a book chapter titledCognitive Interdependence in Close Relationships, finding that long-term couples relied on each other to act as one another's memory banks.

"I really think the internet has become a form of this transactive memory, and I wanted to test it," said Dr Sparrow.

Where, not what

The first part of the team's research was to test whether subjects were "primed" to think about computers and the internet when presented with difficult questions. To do that, the team used what is known as a modified Stroop test.

The standard Stroop test measures how long it takes a participant to read a colour word when the word itself is a different colour - for example, the word "green" written in blue.

Continue reading the main story

“Start Quote

I don't think Google is making us stupid - we're just changing the way that we're remembering things…”

Dr Betsy SparrowColumbia University

Reaction times increase when, instead of colour words, participants are asked to read words about topics they may already be thinking about.

In this way the team showed that, after presenting subjects with tough true/false questions, reaction times to internet-related terms were markedly longer, suggesting that when participants did not know the answer, they were already considering the idea of obtaining it using a computer.

A more telling experiment provided a stream of facts to participants, with half told to file them away in a number of "folders" on a computer, and half told that the facts would be erased.

When asked to remember the facts, those who knew the information would not be available later performed significantly better than those who filed the information away.

But those who expected the information would be available were remarkably good at remembering in which folder they had stored the information.

"This suggests that for the things we can find online, we tend keep it online as far as memory is concerned - we keep it externally stored," Dr Sparrow said.

She explained that the propensity of participants to remember the location of the information, rather than the information itself, is a sign that people are not becoming less able to remember things, but simply organising vast amounts of available information in a more accessible way.

"I don't think Google is making us stupid - we're just changing the way that we're remembering things... If you can find stuff online even while you're walking down the street these days, then the skill to have, the thing to remember, is where to go to find the information. It's just like it would be with people - the skill to have is to remember who to go see about [particular topics]."

Computer ties human as they square off on 'Jeopardy!'

By John D. Sutter, CNN

(CNN) -- The computers haven't proven to be our trivia overlords just yet.
Give them at least until Wednesday.
An IBM supercomputer named Watson finished one round of the TV show "Jeopardy!" on Monday night tied with one of his human competitors and $3,000 ahead of the other.
The man vs. computer face-off won't be complete, however, until the final rounds of the extended trivia game show are aired on Tuesday and Wednesday.
IBM trumpets Watson, which has been in development for years and has the processing power of 2,800 "powerful computers," as a major advancement in machines' efforts to understand human language. The computer receives clues through digital texts and then buzzes in against the two other "Jeopardy!" contestants like any other player would. It juggles dozens of lines of reasoning at once and tries to arrive at a smart answer.
After getting off to a scary-good start, Watson did have a few stumbles.
In one instance, it repeated an answer that another contestant, Ken Jennings, who won 74 "Jeopardy!" episodes in a row, had already tried.
"What is 1920s?" Watson said, sounding like a digitized Matthew Broderick.
"No," game-show host Alex Trebek replied. "Ken said that."
On many other clues, however, Watson was spot-on. After losing the first clue to Brad Rutter, another "Jeopardy!" champion, Watson jumped in on the second question.
Clue: "Iron fitting on the hoof of a horse or a card-dealing box in a casino."
Watson: "What is shoe?"
Correct.
At the start of the show, Trebek went to some lengths to explain the origins of Watson -- IBM approached the show about the idea three years ago -- and how the computer actually works. That's partly because what you see on the "Jeopardy!" stage is somewhat misleading. It looks as if two humans are bookending a simple computer monitor, which appears to be just about as smart as they are. In reality, as Trebek explained, the bulk of Watson's computer power was stored in another building at an IBM lab in New York, where the show is being held for this special three-day competition.
After introducing Watson, to studio applause, this is how Trebek explained it:
"Just as I expected," he said. "That was a very warm reception and I'm sure Watson would have appreciated the applause. Except for one thing: Watson can neither hear nor see. It will be receiving all of its information electronically.
"And as a matter of fact what you're looking at right now is not the real Watson. This is an avatar. This is a representation of Watson. Watson, or course, is a sophisticated computer system too big and too heavy to fit behind that lectern on our stage."
As for the stage version of Watson, his brain-face was represented by a digital Earth that swirled with ribbons of various colors while he thought about questions. As Trebek read the clues, a bar graph appeared at the bottom of the screen, showing the top three answers Watson was considering at that moment and how confident he was in those choices.
Sometimes the computer managed to be confident but still incorrect.
Here's the clue to the first question Watson got wrong:
"From the Latin for end, this is where trains can also originate."
Watson: "What is finis." Confidence level: 97%.
Trebek: "No. Ken?" "What is terminus," Jennings answered correctly.
Before ending the evening tied with Rutter at $5,000 each, Watson had jumped out to an early lead at the first commercial break. At that point, Watson had $5,200 and his closest noncomputer counterpart had only $1,000.
Several Twitter users were awed by the computer's smarts.
"Watson kinda feakin' me out. Big time," Michael Gartenberg, a tech analyst, wrote on his Twitter feed.
Another person wrote: "Watson is almost scary. This is willld! These humans are no match for Watson's algorithms."Trebek summed up the computer's mixed performance this way:

"So, what have we learned so far: Watson's very bright, very fast, but he had some weird little moments once in a while."Then he teased the upcoming shows:"And how many of those will we encounter tomorrow when we play double and final 'Jeopardy!'?"

Man vs. computer: a gaming history

CNN

In 1997, a computer named Deep Blue took a historic victory lap after checkmating world chess champion Garry Kasparov. The IBM computer, capable of processing 100 million board positions a second, was an instant superstar. The win made it less crazy to ask a tantalizing question: Could computers think on their own, and if so, what kind of actions were they capable of?

The word "think" is tricky. Next week, the computer known as Watson will try to beat two "Jeopardy!" champions. Watson is a whiz at math but not at language, so if it wins, a new kind of man vs. machine history will be made. It will show that a computer can dominate at a game that requires reasoning as well encyclopedic knowledge. You can watch Watson in action here.
Years before Kasparov was defeated, in June 1979, computer programmer and chess player Hans J. Berliner's backgammon-playing program beat world champion Luigi Villa 7-1. It is believed to be the first victory by a computer at a game based on strategy, chance and multiple optional positions. Berliner reportedly said that his program wasn't built to analyze millions of moves, like Deep Blue would later, but it computed the benefits and risks of moves.
One of the lesser-known computer victories occurred in the mid-1990s. Marion Tinsley, a math professor and Baptist minister, was the global checkers champ from 1955 to 1992. He lost only seven games in his 45-year career, one of them to a computer in 1994. Called Chinook, the computer was designed by four scientists who worked more than a decade trying to build the perfect checkers dominating program. Tinsley played the machine several times, beat it and then lost in a follow-up match. Chinook went on to beat other humans at the game.
Naturally, the board game of all board games - Scrabble - was next. Don't let its name fool you; Quackle was a formidable Scrabbleist. David Boys, the game's world champion, found that out when the computer beat him in a match in Canada in 2007. But it wasn't like Quackle just walked up to Boys and said, "Let's go." Quackle earned the right to challenge the human only after it defeated another Scrabble program named Maven. Boys was a bit of a sore loser, reportedly telling people that losing to a machine is still better than being a machine.
The same year Quackle won, the first poker game between people and machines involving money was played. A computer project called Polaris, invented at the University of Alberta, beat poker greats Ali Eslami and Phil "The Unabomber" Laak. To be fair, the first time the players faced off against the computer, there was a draw. The computer beat them in the second match. Laak and Eslami brought their A games and won the next two matches.
Ever heard of the game Go? Last summer, a computer beat a Romanian player. The win was remarkable because Go is traditionally challenging for computers. How Stuff Works breaks down why Go is so tough for computers.
But the quest to find out whether man or computer is better at something goes beyond gaming. Computers have been asked to be coaches, design partners, teammates and friends capable of holding conversations with people. In the early 1990s, studies examined computers as social actors, finding that people applied social rules like they were dealing with a person capable of frailties, according to a Stanford University paper summarizing the studies. People even assigned gender to computers based on the sound of the voices coming from the machines, it said.
According to one study, "Individuals can be induced to behave as if computers warranted human treatment, even though users know that the machines do not actually warrant this treatment."
Perhaps that's easier on the ego than knowing a bunch of microchips beat you at the game you play best.