5
$\begingroup$

I am looking for a source that really discusses the classic rules of learning in depth. So classical conditioning, operant conditioning, imitation learning... I have found an infinite number of books that supposedly discuss these topics, but have not yet found anything that summarizes all the important findings on this topic. I am familiar with the basics, but I am explicitly interested in a detailed presentation of the topic.

Can anyone tell me a good source about the different forms of classical learning in mammals? I consider "Reinforcement Learning: An Introduction" comprehensive and detailed regarding RL. A comparable book on biological learning systems would be great. Sources in German and English would fit.

$\endgroup$
  • 1
    $\begingroup$ A "good source" is too subjective here. Please share at least some of the things you found that were not suitable, so that answerers don't find them for you and link them only to be told "I found that already and it was not suitable". Give reasons why not suitable for at least a couple of them. Thanks $\endgroup$ – Neil Slater Mar 13 at 12:04
  • $\begingroup$ I know my request is somewhat vague: I would like a book that presents the basic forms of learning in detail and comprehensibly, but also deals with details. Of course, I am also interested in neuronal correlates where there are insights. What I also notice again and again is that techniques for building complex behavior like chaining are presented and seem simple at first, but in the end I don't know how to implement them in reality. $\endgroup$ – Gurkenkönig Mar 13 at 21:51
  • $\begingroup$ You say you have found an "infinite number" of books. Could you give some specific information about one or two that are close to what you want, and explain what else you want? In your comment you still use vague wording "comprehensive" and "deals with details". Only you understand what you are looking for with this - you need to help readers by giving something more practical for them to work with. $\endgroup$ – Neil Slater Mar 13 at 22:14
  • 1
    $\begingroup$ Alternatively to my first suggestion, if you could reference a book on another subject that you consider comprehensive and detailed, that might help. For instance do you consider "Reinforcement Learning: An Introduction" comprehensive and detailed regarding RL? $\endgroup$ – Neil Slater Mar 13 at 22:15
  • 1
    $\begingroup$ I think there is a book called Neurodynamic programming by Berstekas which is more detailed and probably has some biological facets as well. Also you can consider asking the question in psychology and neuroscience stack exchange. They might have a better idea on the biological facet of learning. $\endgroup$ – DuttaA Mar 17 at 16:26
5
+50
$\begingroup$

I recommend Animal Learning and Cognition: An Introduction by John Pearce.

  • As an introduction to the field, it's comprehensive. It covers all the topics you mentioned and more. Here is a free preview of the book.
  • It's well-regarded, having over 500 citations on Google Scholar.
  • It's relatively modern. Published in 2008, it discusses experimental findings that you won't be able to find in some of the other classics published in the 70s or 80s.

Although it does not specifically focus on mammals, all of the ideas extend to mammals.

$\endgroup$
  • 1
    $\begingroup$ It's a great book with lots of photos and has a dedicated chapter in which is explained how to teach apes the sign language. $\endgroup$ – Manuel Rodriguez Mar 18 at 13:01
  • $\begingroup$ Is this a casual reading book? Just for the sake of increasing your knowledge rather than dabbling into maths? $\endgroup$ – DuttaA Mar 19 at 13:29
  • $\begingroup$ I wouldn't call it casual, but it does have less math than Sutton and Barto's RL book. Keep in mind that we have no complete mathematical model of cognition. This book, along with the rest of the field, is less algorithmic than RL. Instead, it focuses more on experiment and observation. $\endgroup$ – Philip Raeisghasem Mar 19 at 19:08
-1
$\begingroup$

Reinforcement learning for animals asks for the decision making process in lifeforms. A subject has legs, eyes, a skin and a brain and has to take decisions in the environment. For example a dog has to search for food and interpret the language of other dogs.

Sensorimotor learning for mammals can be treated from different perspectives. The first one is biology oriented, in which the brain of real animals is researched. The disadvantage here is, that it is unclear what the neurons in the brain are doing exactly and most of the research is about unsolved problems. If the aim is to get an overview what is known for sure, the better alternative is to focus on animats which are artificial animals. Here is the idea to build in software a simulation of a real animal and put everything what is known about the mammal into the simulation. The subject is called artificial life and has the obligation to animate fishes, dogs and horses on the computerscreen.

A paper which shows a fish (a dolphin is also a mammal) is:

  • Terzopoulos, Demetri, Tamer Rabie, and Radek Grzeszczuk. "Perception and learning in artificial animals." Proc. of the 5th Int. Workshop on Artificial Life: Synthesis and Simulation of Living Systems (ALIFE-96). 1997.
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.