/LMAR=0/XLINE=3/FONT#0=BAXL30/FONT#1=BAXM30/FONT#2=BAXB30/FONT#3=SUB/FONT#4=SUP/FONT#5=BASL35/FONT#6=NGR25/FONT#7=MATH30/FONT#8=FIX25/FONT#9=GRK30



␈↓ ↓H␈↓α␈↓ ¬&THE MINDS OF DOGS

␈↓ ↓H␈↓␈↓ α_One␈α∂of␈α∂the␈α∞major␈α∂areas␈α∂of␈α∞study␈α∂in␈α∂␈↓↓cognology␈↓␈α∞(artificial␈α∂intelligence)␈α∂is␈α∂what␈α∞information
␈↓ ↓H␈↓structures␈α⊂are␈α∂required␈α⊂in␈α⊂order␈α∂to␈α⊂solve␈α⊂different␈α∂kinds␈α⊂of␈α∂problem.␈α⊂ We␈α⊂can␈α∂put␈α⊂this␈α⊂in␈α∂the
␈↓ ↓H␈↓language␈αof␈α␈↓↓common␈αsense␈αpsychology␈↓␈αas␈αthe␈αquestion␈αof␈αwhat␈αkind␈αof␈αmind␈αis␈αrequired␈αto␈αachieve
␈↓ ↓H␈↓certain goals.  In this connection the minds of animals are interesting.

␈↓ ↓H␈↓␈↓ α_Consider the following thought-experiment with dogs:

␈↓ ↓H␈↓␈↓ α_1. There are four dog houses each having a light and a heater.

␈↓ ↓H␈↓␈↓ α_2. These facilities can be operated by a dog using foot switches or nose switches.

␈↓ ↓H␈↓␈↓ α_3.␈α∞The␈α∂dog␈α∞has␈α∂been␈α∞trained␈α∂to␈α∞obey␈α∞commands␈α∂"turn␈α∞on␈α∂the␈α∞light",␈α∂"turn␈α∞off␈α∂the␈α∞light",
␈↓ ↓H␈↓"turn on the heater" "turn off the heater".

␈↓ ↓H␈↓␈↓ α_4.␈α
The␈α
effects␈αof␈α
the␈α
switches␈αin␈α
controlling␈α
the␈αlight␈α
and␈α
heater␈αare␈α
not␈α
constant,␈α
and␈αthe
␈↓ ↓H␈↓dog has learned to operate them until the commanded result has been obtained.

␈↓ ↓H␈↓␈↓ α_5.␈α∂A␈α∂cushion␈α∂is␈α∂fastened␈α∂to␈α∂the␈α∂floor␈α∂of␈α∂one␈α∂of␈α∂the␈α∂houses,␈α∂sometimes␈α∂to␈α∂one␈α⊂house␈α∂and
␈↓ ↓H␈↓sometimes to another.  Other things being equal, the dog favors the house with the cushion.

␈↓ ↓H␈↓␈↓ α_6.␈α
When␈α
it␈α
is␈α
cold,␈α
the␈α
dog␈α
prefers␈α
a␈α
house␈α
with␈α
the␈α
heater␈α
on,␈α
and␈α
when␈α
it␈α
is␈α
hot,␈α
he␈α
prefers
␈↓ ↓H␈↓a house with the heater off.

␈↓ ↓H␈↓␈↓ α_The␈α
null␈α
result␈α
would␈α
be␈α
that␈α
if␈α
the␈α
dog␈α
is␈α
in␈α
the␈α
house␈α
with␈α
the␈α
cushion␈α
and␈α
it␈α
gets␈α
cold,␈α
he
␈↓ ↓H␈↓will go to the warm house rather than turn on the heater in the house with the cushion.

␈↓ ↓H␈↓␈↓ α_The␈αintent␈α
of␈αthe␈α
provision␈αthat␈α
the␈αdog␈α
has␈αbeen␈α
taught␈αto␈α
manipulate␈αthe␈α
switches␈αuntil
␈↓ ↓H␈↓the␈αcommanded␈αcondition␈αhas␈α
been␈αachieved␈αis␈αto␈α
be␈αsure␈αthat␈αthe␈α
dog␈αhas␈αlearned␈αto␈αachieve␈α
the
␈↓ ↓H␈↓result - not merely to perform a specified action.

␈↓ ↓H␈↓␈↓ α_Here are some questions:

␈↓ ↓H␈↓␈↓ α_Supposing␈α∞that␈α
some␈α∞dogs␈α
operate␈α∞heaters␈α
for␈α∞their␈α
comfort,␈α∞here␈α
are␈α∞some␈α∞questions␈α
that
␈↓ ↓H␈↓can be asked:

␈↓ ↓H␈↓␈↓ α_1. Do dogs have preferences about lights?

␈↓ ↓H␈↓␈↓ α_2.␈αDoes␈αtraining␈αa␈αdog␈αto␈αachieve␈αa␈αcondition␈αinsure␈αthat␈αhe␈αwill␈αbring␈αabout␈αthe␈αcondition
␈↓ ↓H␈↓himself␈α⊂when␈α⊂he␈α⊂prefers␈α⊂it?␈α⊂ Is␈α⊂it␈α⊂immediate?␈α⊂ If␈α⊂not,␈α⊂can␈α⊂it␈α⊂be␈α⊂learned␈α⊂in␈α⊂general,␈α⊂so␈α⊂that␈α⊂it
␈↓ ↓H␈↓becomes immediate for a new condition?  If not, can it be learned for each specific condition?

␈↓ ↓H␈↓␈↓ α_3.␈α∞Does␈α∞a␈α∞dog␈α∞learn␈α
something␈α∞in␈α∞order␈α∞to␈α∞please␈α
humans␈α∞more␈α∞readily␈α∞than␈α∞for␈α∞his␈α
own
␈↓ ↓H␈↓benefit?  How about wolves and coyotes?

␈↓ ↓H␈↓␈↓ α_4.␈αSuppose␈αthat␈αa␈αdog␈αhas␈αfirst␈αlearned␈αto␈αrequest␈αa␈αhuman␈αto␈αturn␈αon␈αthe␈αheater␈α(assuming
␈↓ ↓H␈↓that␈αdogs␈αcan␈αlearn␈αthis␈αas␈α
they␈αlearn␈αto␈αrequest␈αto␈αbe␈αlet␈α
in␈αor␈αout).␈α How␈αmuch␈αwill␈αthis␈α
interfere
␈↓ ↓H␈↓with his subsequent learning to turn on the heater for himself?
␈↓ ↓H␈↓␈↓ εH␈↓ 91


␈↓ ↓H␈↓␈↓ α_One␈αmay␈αconjecture␈αthat␈αliving␈αwith␈αhumans␈αinterferes␈αwith␈αdogs'␈αlearning␈αto␈αdo␈αthings␈αfor
␈↓ ↓H␈↓themselves.␈α Imagine␈αthat␈αwe␈αlived␈αin␈αclose␈αproximity␈αwith␈αgods␈αwho␈αdid␈αhighly␈αvisible␈αthings␈αfor
␈↓ ↓H␈↓totally␈α∩inexplicable␈α∩reasons␈α∩and␈α∩sometimes␈α∩rewarded␈α∩and␈α∩punished␈α∩us␈α∩for␈α∩reasons␈α∪that␈α∩were
␈↓ ↓H␈↓sometimes␈αexplicable␈αand␈αsometimes␈αwere␈αnot.␈α It␈αseems␈αthat␈αit␈αwould␈αinterfere␈αwith␈αour␈αability␈αto
␈↓ ↓H␈↓learn␈α
to␈α
achieve␈α
goals.␈α
 In␈α
fact,␈α
we␈α
can␈α
perhaps␈α
characterize␈α
the␈α
original␈α
form␈α
of␈α
religion␈α
as␈αthe
␈↓ ↓H␈↓hypothesis␈α∂that␈α∞there␈α∂were␈α∂such␈α∞gods,␈α∂and␈α∂speculate␈α∞that␈α∂the␈α∂hypothesis␈α∞did␈α∂interfere␈α∂with␈α∞our
␈↓ ↓H␈↓acquiring competence in understanding and controlling our environment.

␈↓ ↓H␈↓␈↓ α_On␈α⊂this␈α∂basis,␈α⊂we␈α∂might␈α⊂conjecture␈α⊂that␈α∂dogs␈α⊂or␈α∂apes␈α⊂would␈α∂develop␈α⊂a␈α⊂more␈α∂interesting
␈↓ ↓H␈↓culture␈α
if␈α
they␈α
lived␈α
in␈α
total␈α
ignorance␈αof␈α
humans,␈α
and␈α
if␈α
their␈α
enviroment␈α
had␈α
many␈αuseful␈α
causal
␈↓ ↓H␈↓relations␈α∂and␈α∞that␈α∂apparent␈α∂miracles,␈α∞e.g.␈α∂a␈α∞new␈α∂object␈α∂putting␈α∞in␈α∂an␈α∞appearance␈α∂one␈α∂fine␈α∞day,
␈↓ ↓H␈↓happened␈αrarely␈αif␈αat␈αall.␈α This␈αmay␈αbe␈αrequired␈αfor␈αapes␈αto␈αreach␈αtheir␈αfull␈αintellectual␈αpotential.
␈↓ ↓H␈↓Perhaps␈αa␈αcompromise␈αcould␈αbe␈αdeveloped␈αthat␈αwould␈αstill␈αallow␈αapes␈αto␈αbe␈αtaught␈αsign␈αlanguage,
␈↓ ↓H␈↓but␈αwould␈αminimize␈αmiracles␈α
in␈αtheir␈αenvironment.␈α The␈α
best␈αenvironment␈αmight␈αnot␈α
be␈αnatural.
␈↓ ↓H␈↓A␈α∞simpler␈α∞environment␈α∞of␈α
geometrically␈α∞shaped␈α∞simply␈α∞colored␈α
rooms␈α∞might␈α∞be␈α∞better,␈α∞but␈α
once
␈↓ ↓H␈↓the apes are introduced to it, it shouldn't be changed except by comprehensible processes.

␈↓ ↓H␈↓John McCarthy

␈↓ ↓H␈↓The source of this is DOGMIN[E77,JMC] at SU-AI.  Version pubbed at 12:56.