Brain Science Podcast discussion

How We Learn: Why Brains Learn Better Than Any Machine . . . for Now
This topic is about How We Learn
22 views
2020 > BS 167 with Stansislas Dehaene

Comments Showing 1-5 of 5 (5 new)    post a comment »
dateDown arrow    newest »

Ginger Campbell (gingercampbell) | 321 comments Mod
I have just posted BS 167 with Stanislas Dehaene. Dehaene is a French neuroscientist who has published numerous books about his work. In this interview we talk about his newest book How We Learn: Why Brains Learn Better Than Any Machine . . . for Now. It is a great combination of cool science and practical information for teachers, parents, and life-long learners.

Listen now

Learn more.


message 2: by Dalton (new)

Dalton | 16 comments Bayesian Statistics -

Bayesian Statistics provides us with mathematical tools to update our beliefs about random events in light of seeing new data or evidence about those events.

First, I'll admit to a prejudice and dislike for the use of statistics and in particular, Bayesian Statistics - especially in AI. As can be seen from the above definition, Bayesian statistics recognizes the inherent limitations of statistical accuracy. The problem stems from human implementation, not necessarily the principles embodied in statistics.

As it may apply to cognition, I don't see why they resort to supplanting the age old Trial and Error or if you don't succeed, try try again.

The application of Bayesian Statistics in AI (deep learning) is a mathematician's approach to avoiding the work that would be necessary to properly include Context as part of its procedures for interpretation. Instead of implementing the influence of context to make correct interpretations, they simply count instances and select the most common or popular usage as a foundation for interpretation. The brain is not Bayesian.


message 3: by Dalton (new)

Dalton | 16 comments Learning and the effects of Pruning -

It's well established that pruning will eventually limit certain kinds of learning. The window in which people learn and recognize some foreign phonemes are an example. The difficulty in learning that people encounter as they reach adulthood has more to do with the ease of resource allocations. The adolescent brain is slow to establish all its inhibitory connections (lateral, configuration, and volitional), thus allowing the imprinting of concepts to be far easier. The establishment of inhibitory connections impede new learning for the sake of stability and the prevention of chaos/confusion.

Also, the effects of novelty on focus and attention drop off as analogically similar experiences become common and sufficiently close to other past experiences. Habituation and the redundancy of experiences tend to limit exposure to novel experiences as well and thereby slow down learning.

I suspect that the demands on learning, and the complexity of what's attempted in adulthood, are also responsible for the difficulty experienced. The school years are dominated by rote learning or learning without meaning. Later in life, we demand depth to what we learn and that requires assimilation. Assimilation requires the linking of conceptual relationships, the foundation of meaning. Linking requires the growth of axonal projections and the establishment of new synapses with their targets. Growth takes time.


message 4: by Dalton (new)

Dalton | 16 comments Sleep - Consolidation -

I really don't think that much in the way of rehearsal occurs during sleep. I believe that consolidation occurs as the result of biochemical changes that activate synapses. After sufficient stimulation of the synapse (LTP), subcellular deficits feedback and activate the genetic production of enzymes required to modify the synapse - activating it long term. This all takes time and energy. Sleep relieves a large amount of drain on energy resources (ATP) that can be drafted during sleep to drive the processes needed to instantiate and maintain those new synapses. Support for this comes from Kendall et al in which they established the principles of synaptic activation long term and the maintenance by enzymes that have prion segments


message 5: by Yon (new)

Yon - | 2 comments This was a great interview about a GREAT book!!!

The coolest idea for me: that we reuse brain regions that have the right basic mechanisms for new skills, for example, we take over face recognition regions to use for letter recognition.

Also, a great quote by Dehaene in the interview "Consciousness is just the tip of the iceberg"


back to top