rec_learning_function: rec_learning_function
In KennethEnevoldsen/SiRToM: SirToM: Simulation in R: Theory of Mind.

k-ToM's learning function, where it updates its level probability, parameter, choice probability and gradient etimates. This is called recursively.

1 2	rec_learning_function(prev_hidden_states, params, choices, level, player, p_matrix)

`prev_hidden_states`	a list structure containing the states from last round
`params`	a list structure containing k-ToM's volatility parameter
`choices`	a vector of 1 own choice and 2 opponent's choice from last round
`level`	k-ToM's sophistication level
`player`	k-ToM's player role, i.e. which side of the payoff matrix is used
`p_matrix`	a given 2-by-2 payoff matrix