In the courses, we are taught economics theoretically,inclusive of assumptions,simplicity and inferences based on the optimal thinking. But, later on we see that most of the theories and hypothesis aren't reflective of the ground reality. Most peoples are under the thread of cognitive bias, but in the courses we are taught they are Rational. As cited by Mr. Richard Thaler that, we are Humans not ECONS. Then why is it that without delving deep into the realistic part we start with the economics that is all theory.It's not that they aren't important,they are the Bible to build foundation and concept, but still,even without it can't the economics that is real be taught directly?(with foundations and concepts included?)
The main reason answers to the question "what is the difference between normative and positive economics?". Under the normative approach, we have a benchmark for what "should" happen if we all satisfied some theoretical assumptions. Under this logic, anything that you observe in the real world that deviates from the theoretical norm can be traced to failure of your assumptions holding in the real world (e.g. concepts which assume equilibrium in the long run, etc). In other words the theory is good if the assumptions are good, but it doesn't apply in reality because the assumptions are not met.
Well, this is at least how we always thought until around 1953 with Milton Friedman's famous book "Essays in Positive Economics" where he roughly argued that economists should stop making unrealistic assumptions because the test of any scientific economic model should solely be based on whether it predicts well or not. He implied that wrong assumptions could make good theories as well, as long as those theories had a strong empirical backbone. So fast forward, there was this epistemological revolution (I'll let you search for Karl Popper etc) where our basis for judging "good" theories changed.
When it comes to behavioural economics, you have to understand that it started from different places. On one hand, you have some guys who advocated for relaxing rationality assumptions (Herbert Simon considered that rationality of agents was bounded) and considering cognitive biases (Daniel Kahneman and Amos Tversky) in decision theory (which ultimately implies abandoning von neumann and morgernstern's expected utility framework). On the other hand, some guys were using experiments to test standard economic theories and found a lot of hypotheses not matching with their findings.
There are a lot more things which happened, but these distinct (each on a separate island) agendas created more and more interest until becoming orthodox now.Part of why it became orthodox is because now there seems to be a good body of models that have better predictions.
So should we abandon standard economic theories? NO!! Because as I mentioned, we are only part of an epistemological revolution. This does not in any case mean that we are close to the truth (Proof is the recent "replication crisis"). Moreover, these standard economic models are not false in themselves. Empirical data still show that they can make accurate predictions.
Point is that, even if we only did behavioural economics, in the long run, our theories would converge to what has already been demonstrated using standard economic models.
Finally, to be able to teach both standard economic models and behavioural economics simultaneously would take too much time for any average student to ultimately become good in any of them; by trying to be expert in both, one becomes expert in nothing.
Behavioral economics is in large parts a collection of models and theories that purport to explain deviations from the predictions of neo-classical theory. For this reason, textbooks in behavioral economics include neo-classical theory, experimental evidence of deviations, and models trying to explain those. (This is why the Dhami has 1764 pages.) To understand this field one has to understand the predictions of neo-classical theory first. Therefore it makes sense to start by teaching neo-classical theory.
Another reason is that many behavioral models are explanatory but hardly predictive, i.e. they often contain several parameters that have to be estimated before the model can be applied, and then they typically make superior predictions only in a very narrow range. If you need easy predictions about many different economic interactions, then by and large, neo-classical theory still performs best.