<s>
FORR	B-Architecture
(	O
FOr	B-Architecture
the	I-Architecture
Right	I-Architecture
Reasons	I-Architecture
)	O
is	O
a	O
cognitive	B-Architecture
architecture	I-Architecture
for	O
learning	O
and	O
problem	O
solving	O
inspired	O
by	O
Herbert	O
A	O
.	O
Simon	O
's	O
ideas	O
of	O
bounded	O
rationality	O
and	O
satisficing	O
.	O
</s>
<s>
It	O
has	O
been	O
used	O
in	O
game	B-Algorithm
playing	I-Algorithm
,	O
robot	O
pathfinding	O
,	O
recreational	O
park	O
design	O
,	O
spoken	B-General_Concept
dialog	I-General_Concept
systems	I-General_Concept
,	O
and	O
solving	O
NP-hard	O
constraint	B-Application
satisfaction	I-Application
problems	I-Application
,	O
and	O
is	O
general	O
enough	O
for	O
many	O
problem	O
solving	O
applications	O
.	O
</s>
<s>
FORR	B-Architecture
does	O
not	O
have	O
perfect	O
knowledge	O
of	O
how	O
to	O
solve	O
a	O
problem	O
,	O
but	O
instead	O
learns	O
from	O
experience	O
.	O
</s>
<s>
Intelligent	B-General_Concept
agents	I-General_Concept
are	O
not	O
optimal	O
,	O
but	O
make	O
decisions	O
based	O
on	O
only	O
a	O
subset	O
of	O
all	O
possible	O
good	O
reasons	O
and	O
informative	O
data	O
.	O
</s>
<s>
This	O
idea	O
of	O
bounded	O
rationality	O
was	O
introduced	O
by	O
Herbert	O
A	O
.	O
Simon	O
,	O
who	O
along	O
with	O
Allen	O
Newell	O
developed	O
the	O
early	O
foundations	O
of	O
the	O
study	O
of	O
cognitive	B-Architecture
architectures	I-Architecture
and	O
also	O
inspired	O
early	O
architectures	O
such	O
as	O
Soar	B-Architecture
and	O
ACT-R	B-Language
.	O
</s>
<s>
FORR	B-Architecture
depends	O
upon	O
the	O
idea	O
that	O
there	O
are	O
multiple	O
reasons	O
or	O
rationales	O
for	O
performing	O
actions	O
while	O
solving	O
a	O
problem	O
.	O
</s>
<s>
The	O
sometimes-right	O
reasons	O
can	O
compete	O
with	O
each	O
other	O
:	O
for	O
example	O
,	O
in	O
game	B-Algorithm
playing	I-Algorithm
,	O
one	O
good	O
reason	O
might	O
be	O
to	O
capture	O
pieces	O
,	O
while	O
another	O
might	O
be	O
to	O
control	O
some	O
area	O
of	O
the	O
board	O
.	O
</s>
<s>
In	O
FORR	B-Architecture
,	O
these	O
competing	O
reasons	O
are	O
called	O
Advisors	O
.	O
</s>
<s>
Because	O
of	O
its	O
reliance	O
on	O
a	O
set	O
of	O
independent	O
agents	O
(	O
the	O
Advisors	O
)	O
,	O
FORR	B-Architecture
can	O
be	O
considered	O
a	O
connectionist	O
architecture	O
.	O
</s>
<s>
A	O
FORR	B-Architecture
architecture	O
has	O
three	O
components	O
:	O
a	O
set	O
of	O
descriptives	O
that	O
describe	O
the	O
state	O
of	O
the	O
problem	O
,	O
a	O
tiered	O
set	O
of	O
Advisors	O
that	O
are	O
consulted	O
in	O
order	O
to	O
decide	O
what	O
action	O
to	O
perform	O
,	O
and	O
a	O
behavioral	O
script	O
that	O
queries	O
the	O
Advisors	O
and	O
performs	O
the	O
action	O
that	O
they	O
suggest	O
.	O
</s>
<s>
Once	O
the	O
problem	O
domain	O
is	O
identified	O
,	O
the	O
implementation	O
of	O
a	O
FORR	B-Architecture
architecture	O
for	O
that	O
domain	O
has	O
two	O
basic	O
stages	O
:	O
finding	O
possible	O
right	O
reasons	O
(	O
the	O
Advisors	O
)	O
and	O
learning	O
their	O
weights	O
for	O
a	O
particular	O
class	O
.	O
</s>
<s>
FORR	B-Architecture
learns	O
the	O
weights	O
for	O
its	O
tier	O
3	O
Advisors	O
by	O
experience	O
.	O
</s>
<s>
FORR	B-Architecture
has	O
been	O
used	O
for	O
game	B-Algorithm
playing	I-Algorithm
,	O
robot	O
pathfinding	O
,	O
constraint	B-Application
satisfaction	I-Application
problems	I-Application
,	O
park	O
design	O
,	O
and	O
spoken	B-General_Concept
dialog	I-General_Concept
systems	I-General_Concept
.	O
</s>
