dim_codings.txt

Task1:
------
The original stimuli (8) have 5 features. In Mack's experiments, 3 features are used.
In the files, a stimulus is named something like 12111 where each number corresponds to 
one dimension being 1 or 2. For example, the first dim corresponds to legs.
Here, we map out the index of three dims that Mack used and we will be using.

1st dim = legs
3rd dim = antenna
4th dim = mouth

So if we ignore the other two dims, the files should be named as:
111
112
121
122
211
212
221
222

if we convert to binary:
000
001
010
011
100
101
110
111

So in hash table, if would be:
dict = {'12111': [0, 0, 0], 
        '12121': [0, 0, 1],
        '12211': [0, 1, 0],
        '12221': [0, 1, 1],
        '22111': [1, 0, 0],
        '22121': [1, 0, 1],
        '22211': [1, 1, 0],
        '22221': [1, 1, 1]}
This is so we can train a model with binary coding as labels.

Task2-5:
--------
Things are a lot easier for these tasks because the 3 dimensions were explicitly binary coded.

Task6:
------
Exception here where there are 4 features and not binary coded when provided.
Here we explicitly map out the conversion:

cgll - 0000
cglr - 0001
cgsl - 0010
cgsr - 0011
crll - 0100
crlr - 0101
crsl - 0110
crsr - 0111

tgll - 1000
tglr - 1001
tgsl - 1010
tgsr - 1011
trll - 1100
trlr - 1101
trsl - 1110
trsr - 1111