Dolphins Have Names, Demonstrate Capacity for Learned Language in the Wild
They seemed to be using the whistles to keep in touch with the dolphins they knew best, just as two friends might if suddenly and unexpectedly separated while walking down a street.
Moreover, copying wasn’t exact, but involved modulations at the beginning and end of each call, perhaps allowing dolphins to communicate additional information, such as the copier’s own identity.
That possibility hints at what linguists call referential communication with learned signals, or the use of learned rather than instinctively understood sounds to mentally represent other objects and individuals. As of now, only humans are known to do this naturally.
“We learn language and refer to objects. This has been shown with captive dolphins and captive gray parrots, but hasn’t been seen in the natural communication system of any species,” said King. “We’re not saying that this is what they’re doing, but we’re definitely suggesting that we should look into it.”
(via Dolphins May Call Each Other by Name | Wired Science | Wired.com)
Robo Grading: Another Reason to Send Your Kids to Private School
Since 2009, Utah has used computers to grade essays on a state student-assessment test. And testing companies use essay-evaluating software as one of two graders on graduate-school admissions exams such as the GRE.
But how well, really, can a computer grade an essay? To find out, Mark Shermis, an education researcher at the University of Akron, ran 22,029 standardized middle- and high-school essays through software from eight companies (plus one open-source algorithm).
The programs, which generally track content, organization and style, generated results indistinguishable from those of humans—just much faster. With that kind of efficiency, robot graders could mean more homework for students everywhere.
If AI grades papers similarly to human teachers, it’s hard to imagine that cash-strapped public schools won’t adopt robo-grading en masse.
What’s missing in these kinds of studies is the reality that a human teacher reads and grades an essay and then returns to the classroom and works with the student to improve, based on what she saw in the paper she graded.
Unfortunately, the public school of the future is likely to be rows and rows of students sitting in front of monitors, working on “customized” lesson plans, generated and evaluated by computers, with one or more adult proctors maintaining discipline.
This means, in wealthy communities there will be more, and better trained human teachers in smaller classrooms using newer technology. Conversely in poorer communities there will be fewer teachers - glorified corrections officers - working in huge rooms full of old, poorly maintained computers with out-of-date software.
Only kids whose parents can afford to send them to private school will get real teaching and real human interaction during the school day.
(via Robo-Grading Programs Judge Student Essays Better Than Humans Do | Popular Science)
Dogs Process Language At The Level of a 2-3 Year-Old Child
To acquire language, kids use a strategy called “fast mapping” — forming quick, rough hypotheses about the meaning of new words after just one or two exposures. So do dogs.
Recently, researchers found that a border collie named Rico was able to infer the names of more than 200 items using this method. Four weeks after the initial exposure, Rico was still able to retrieve the items by name.
Another border collie in South Carolina has memorized over 1,000 nouns. The dog, Chaser, reportedly loves her vocabulary drills.
Researchers Affect Memory Consolidation During Deep Sleep, Improving Learning
In the Northwestern study, research participants learned how to play two artificially generated musical tunes with well-timed key presses. Then while the participants took a 90-minute nap, the researchers presented one of the tunes that had been practiced, but not the other.
“Our results extend prior research by showing that external stimulation during sleep can influence a complex skill,” said Ken A. Paller, professor of psychology in the Weinberg College of Arts and Sciences at Northwestern and senior author of the study.
By using EEG methods to record the brain’s electrical activity, the researchers ensured that the soft musical cues were presented during slow-wave sleep (deep sleep, not REM sleep, or dreaming), a stage of sleep previously linked to cementing memories.
Participants made fewer errors when pressing the keys to produce a melody that had been presented while they slept, compared to the melody not presented.
“We also found that electrophysiological signals during sleep correlated with the extent to which memory improved,” said lead author James Antony of the Interdepartmental Neuroscience Program at Northwestern. “These signals may thus be measuring the brain events that produce memory improvement during sleep.”
(via How to reinforce learning while you sleep | KurzweilAI)
There is a growing body of scholarly research suggesting that, when used properly, social media can boost both learning outcomes and student engagement. The key phrase in that sentence is “when used properly.” The problem is that research in this area is still relatively limited, and most of what is being done in classrooms is experimental. No one has figured out definitively what does and does not work.
Tutoring Software Senses Human Emotions, Adapts Methods to Match
“AutoTutor” and “Affective AutoTutor” can gauge the student’s level of knowledge by asking probing questions, analyzing the student’s responses to those questions; proactively identifying and correcting misconceptions; responding to the student’s own questions, gripes, and comments; and even sensing a student’s frustration or boredom through facial expression and body posture and dynamically changing its strategies to help the student conquer those negative emotions.
“Most of the 20th-century systems required humans to communicate with computers through windows, icons, menus, and pointing devices,” says D’Mello, who specializes in human-computer interaction and artificial intelligence in education. “But humans have always communicated with each other through speech and a host of non-verbal cues such as facial expressions, eye contact, posture, and gesture. In addition to enhancing the content of the message, the new technology provides information regarding the cognitive states, motivation levels, and social dynamics of the students.”
Japanese Robot Can Learn, Act Autonomously
When compared to similar robots out there, “HIRO” is pretty impressive. As you can see in the video embedded below, the robot uses a algorithm called SOINN (Self-Organizing Incremental Neural Network) [JP] to do its magic…
When confronted with a new problem, HIRO tries to “remember” past actions and accesses stored information in order to solve it. It also can be “taught” to do certain tasks, and it’s able to learn and react by analyzing the world around him.
(via Robot HIRO Learns, Adapts To New Situations Autonomously (Video) | TechCrunch)
AI Demonstrates Natural Learning, Applies New Skills To Civilization
At first the AIs just interacted randomly. But then they gave one AI access to the user’s manual. It used its ability to read text from the screen to correlate actions with words, and words with items from the manual. Pretty soon it was winning 79% of its games — completely based on its own natural learning processes. …The “naive” AIs in the MIT experiment could only build up a knowledge base from scratch.
The deeper purpose of the experiment was to demonstrate that AIs could determine the meanings of words by interacting with their environment. Today Civilization, tomorrow a factory or hospital, where a drone of some sort might learn new words and actions just from hearing them and seeing them done.
