Numpy argmax function error

Question

I am trying to use the argmax function of numpy to take two argument. Here is the code:

UP = 1 
DOWN = 2 
LEFT = 3 
RIGHT = 4 
actlist = [UP, DOWN, LEFT, RIGHT]

a = numpy.argmax(actlist, lambda a: expected_utility(a,i,j,U))

The problem is, one of the arguments is the return value of a function. It returns an integer.
The function expected_utility(a,i,j,U) looks like:

def expected_utility(a,i,j,U):
    return sum(p*U[i][j] for (k,l,p) in T(i,j,a))

If I use lambda with the second parameter of the argmax function, I am getting an error:

TypeError: 'function' object cannot be interpreted as an integer

If I don't use lambda with the second parameter of the argmax function, I am getting the error:

UnboundLocalError: local variable 'a' referenced before assignment

I am using Python 3.6.5

Now I am even confused with how argmax function even works. What is the role of lambda here.