Where Can I Buy Toblerone?

Learning Levels

One of the most important topics of neuroscience and learning, is a crucial challenge to understand what algorithms that enable learning. And, of course, if these algorithms are real in the sense of the term Thuringia.

Oovviamente we are not referring only to learn at school, or, more generally, in the human realm. We refer instead to a much broader concept. Now we examine some of the levels at which learning can take place.

Small premise: these are just my personal opinions on the subject, and I will not say anything definitive, or even genuinely scientific, matter.

Learning explicit heteronomous: This indicates the learning they are able to complete living if trained. This covered a large part of education human, but also the training of dogs and horses by the man; education for hunting animals. The transmission, in short, non-innate behavior among different individuals. The culture in the broadest sense, if you like.

Learning implicit self: every living being, when exposed to environmental pressure, changing its behavior to adapt to the environment. This applies to animals with higher intellectual capacity on average, say, mammals and birds, for animals with lower powers, such as fish or insects. Let me give two examples that struck me very much.

zoo in Stuttgart, there are many areas for animals separated by moats, rather than cages. In these ditches live carp. If you get close to observe the animals casting a shadow on fossatto observe that the carp come to the surface, opening her mouth. Evidently they have learned that often the visitors throw them food.

At a still more primitive: the C. Elegans stores during the early stages of development to which the temperature was raised, and always tends to move towards similar temperatures.

The example of C. Elegans is very extreme! In fact, our nematode has a few hundred neurons, so let's see how the behavior of movement must be somehow encoded within this small neural network, using the its connections with the sensory neurons. This brings us to the learning of lower level.

Parametric Learning: This is what I do neural networks, a neural network output is rewarded or not rewarded with a (virtual) and, according to this reward, are reinforced or modified connection, so that the network tends to maximize the reward you get. This is the principle of reinforcement learning . Obviously, for this type of learning is necessary that it provides a higher reward.

What are the great challenges of neuroscience? Let's see some:

- understand the exact mechanisms of learning Parametric;
- What are rewarding instances of the various circuits;
- understand which and how many levels of learning involved and how they are intertwined;
- you can understand how the 'emergency an explicit learning higher level.

understand the difficulties?