Language - Computational Explorations in Cognitive Neuroscience - page 330

trial
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39

Event
tart_tttartt
tent_tttentt
face_fffAsss
deer_dddErrr
coat_kkkOttt
grin_grrinnn
lock_lllakkk
rope_rrrOppp
hare_hhhArrr
lass_lll@sss
flan_fllonnn
hind_hhhIndd
wave_wwwAvvv
flea_fllE−−−
star_sttarrr
reed_rrrEddd
loon_lllUnnn
case_kkkAsss
flag_fll@ggg
post_pppOstt
tact_ttt@ktt
rent_rrrentt
fact_fff@ktt
deed_dddEddd
cost_kkkostt
gain_gggAnnn
lack_lll@kkk
role_rrrOlll
hire_hhhIrrr
loss_lllosss
plan_pll@nnn
hint_hhhintt
wage_wwwAjjj
plea_pllE−−−
stay_sttA−−−
need_nnnEddd
loan_lllOnnn
ease_−−−Ezzz
flaw_fllo−−−
past_ppp@stt

min_Phon_dist ev_nm sm_nm con_vis con_sem

0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

abs_vis
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
1
0
0
0
0

abs_visem
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0

abs_sem
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
1
1
0
0
0
0
0
0
0
1
0

0 grin_grrinnn 1 0

0 tent_tttentt 0 0

0 face_fffAsss 0 0

0 deer_dddErrr 0 0

0 rope_rrrOppp 1 0

0 stay_sttA−−− 1 0

0 loan_lllOnnn 1 1

0 tact_ttt@ktt 1 0

0 tart_tttartt 1 1

0 star_sttarrr 1 0

0 loan_lllOnnn 1 1

0 hind_hhhIndd 0 0

0 case_kkkAsss 1 1

0 tact_ttt@ktt 1 0

1.86154 flag_fll@ggg 1 0

0 reed_rrrEddd 0 0

0 loon_lllUnnn 0 0

0 flag_fll@ggg 1 0

0 flag_fll@ggg 0 0

0 rope_rrrOppp 1 0

0 tact_ttt@ktt 0 0

0 rent_rrrentt 0 0

0 hint_hhhintt 1 0

0 deed_dddEddd 0 0

0 cost_kkkostt 0 0

0 gain_gggAnnn 0 0

0 need_nnnEddd 1 0

0 rent_rrrentt 1 0

0 hire_hhhIrrr 0 0

0 cost_kkkostt 1 0

0 need_nnnEddd 1 0

0 hint_hhhintt 0 0

0 wage_wwwAjjj 0 0

0 plea_pllE−−− 0 0

0 stay_sttA−−− 0 0

0 reed_rrrEddd 1 0

0 loan_lllOnnn 0 0

0 lock_lllakkk 1 0

0 need_nnnEddd 1 0

0 past_ppp@stt 0 0

Figure 10.9: Text log output from complete direct pathway lesion, showing selected columns with relevant data. In addition to
those from the previous figure, con sem and abs sem show semantic errors for concrete and abstract, and abs visem shows
visual and semantic errors for abstract words. There were also blend and other errors that are not shown.

“common sense” judgment and the cluster plot in fig-
ure 10.7 (which usually agrees with common sense, but
not always) to determine if there is a semantic similarity
between the response and the input word (i.e., semantic
errors) — count the number of cases for which you think
this is true.

tween the patterns, then errors are scored as semantic.
The formula for the cosine is:

(10.1)

which goes from 0 for totally non-overlapping patterns
to 1 for completely overlapping ones. The value of
.4 does a good job of including just the nearest neigh-
bors in the cluster plot shown in figure 10.7. Neverthe-
less, because of the limited semantics, the automatically
coded semantic errors do not always agree with our in-
tuitions.

The simulation also does automatic coding of seman-
tic errors, but they are somewhat more difficult to code
because of the variable amount of activity in each pat-
tern. We use the criterion that if the input and response
semantic representations overlap by .4 or more as mea-
suredbythe cosine or normalized inner product be-