How We Learn

Paraplegics can “learn” to walk again with the assistance of virtual reality technology that takes advantage of the brain’s ability to change itself. Consult further: “This is the first fMRI study in the literature that provides evidence for neuroplasticity and associated locomotor recovery after VR.”

↩︎ posted Oct. 6, 2016

Lumosity Dims

“Brain training” app Lumosity is reeling, with app downloads down and thousands of refund requests reaching their inbox after the government ruled their advertisements were unsupported by science, with no evidence that the skills these games impart are exportable to general intelligence.

The industry is big, and expected to grow to be $3.4 billion by 2020. But Lumosity’s problem isn’t isolated. Psychologists reviewed every paper put forward by the brain training industry, and all were found wanting on similar grounds.

↩︎ posted Oct. 5, 2016

Adderall is at the center of an “epidemic of over-diagnosis and addiction.” Its fate has been linked to Big Pharma’s pockets from the get-go, but who it’s prescribed to has changed. Best read if you have the time: a journal paper on the neuroethics of Adderall and other memory-enhancing drugs making their way from schools into workplaces, including professional sports.

↩︎ posted Oct. 5, 2016

The Benefits of Deep Learning

On the one hand, no, we shouldn’t go to the lengths of rote learning in Pakistan. But Common Core might have overemphasized the conceptual side of learning, minimizing the need for a solid base of experience.

From a woman who taught herself math and engineering late in life: “The problem with focusing relentlessly on understanding is that math and science students can often grasp essentials of an important idea, but this understanding can quickly slip away without consolidation through practice and repetition. Worse, students often believe they understand something when, in fact, they don’t.”

In other words, deep learning is as important for human intelligence as it is for AI.

↩︎ posted Oct. 5, 2016

The Limits of Neural Nets

Neural networks modeled on theories of emergent learning and decision-making processes in the brain are becoming a near-ubiquitous tool in modern science, used for everything from cracking Go to recognizing gestures. But while they’re good at identifying correlations, neural nets aren’t that great at actually explaining connections. That can lead to thorny ethics in their application:

Imagine that in the near future, a deep-learning neural network is trained using old mammograms that have been labelled according to which women went on to develop breast cancer. After this training, the tissue of an apparently healthy woman could already ‘look’ cancerous to the machine. “The neural network could have implicitly learned to recognize markers — features that we don’t know about, but that are predictive of cancer,” he says.

But if the machine could not explain how it knew. It would present physicians and their patients with serious dilemmas.

↩︎ posted Oct. 6, 2016