Why is this method of copying a list so much faster than these others?

I have run this list copying experiment and found one of the times to be much quicker than the others, as far as I can tell it does copy the list, unless someone can explain otherwise?

import time

def test(d):
    s = time.time()
    for i in range(20000000):
        d['b'] = d['a'].copy()
    e = time.time()
    print (".copy(): "+str(e-s))

    s = time.time()
    for i in range(20000000):
        d['b'][:] = d['a']
    e = time.time()
    print ("[:] on left: "+str(e-s))

    s = time.time()
    for i in range(20000000):
        d['b'] = d['a'][:]
    e = time.time()
    print ("[:] on right: "+str(e-s))

d = {'a': [1,2,3,4],
     'b': []}
test(d)

results:

.copy(): 4.150316476821899
[:] on left: 2.535377025604248
[:] on right: 4.140159606933594

To prove [:] on left is copying?

d['b'][:] = d['a']
del d['a'][1]
d['a'][0] = "testing"
print (d)

prints:

{'a': ['testing', 3, 4], 'b': [1, 2, 3, 4]}

Can anyone explain what's going on here?