I would agree. 
I call this the deliberation loop versus the reaction loop.  The reaction loop 
deals with familiar situations(habitual and reflex responses).  The 
deliberation loop deals with situations requiring planning.  Theseloops are 
detailed in my paper "Patterns for Cognitive Systems".
Cheers, 
~PM.

Date: Tue, 4 Dec 2012 18:11:48 -0600
Subject: Re: [agi] Kurzweil's new book
From: [email protected]
To: [email protected]

I don't think rational vs. irrational really maps to conscious vs. 
subconscious. A rote behavior can still be an intelligent/rational one, but it 
is intelligence that has been automated. If consciousness is access to 
information by the part of the brain which handles the unexpected, then we 
should expect to be less conscious of habitualized tasks than of novel/creative 
ones. This reflects what I think is a clear distinction in the architecture of 
the mind between fluid and crystallized intelligence (as Todor already pointed 
out).

>From this perspective, the conscious mind has the specific purpose of dealing 
>with novelty, and the subconscious mind has the specific purpose of dealing 
>with familiarity. And so when we look to building an AGI, we should expect to 
>build a "subconscious" subsystem which handles familiar situations via 
>feature-based lookup of an appropriate decision procedure, and a "conscious" 
>subsystem which manages the "subconscious" subsystem by using situational 
>analysis to add dynamically generated new decision procedures or improve 
>existing decision procedures. The "conscious" subsystem acts as a default 
>handler of last resort for situations that aren't already covered by the 
>"subconscious" subsystem, and is far more general in its capabilities, at the 
>expense of being much slower and less reliable in its worst-case performance. 
>(I am reminded of the classic exploration/exploitation dichotomy in 
>reinforcement learning algorithms.) Both subsystems, however, are mostly 
>rational, and when either fails we call it irrationality.


On Sun, Dec 2, 2012 at 7:12 AM, Todor Arnaudov <[email protected]> wrote:

Hi Ben, Mike,
Ben>Whoa... that's a kinda surreal perspective!!!

Ben>The unconscious is rational and algorithmic?? Â  That would be a big 
surprise to the psychiatry community ;O ;D ...

Ben>Mike T, thanks for being baffling and silly in a different way than your 
usual; that brightened up a foggy, rainy Hong Kong morning for me ;)


On Sun, Dec 2, 2012 at 8:42 AM, Mike Tintner <[email protected]> wrote:




Yes. Put that more simply, the conscious mind supervises creative thinking -Â  
that which “we don’t know how to do†  pace Piaget, and wh. is 
non-algorithmic,  - and the unconscious mind is in charge of routine, 
(basically rational), algorithmic thinking, which we do already know how to do. 
And that’s the essential architecture of a mildly evolved AGI or lower 
organism – and a neat, fairly obvious division of labour.

 



Ben, in fact, yes, the "unconscious" mind is *obviously* in charge of routine 
thinking, and it is also "rational", at least in my definitions of it, I wrote 
about also in the publications I linked in the previous email. I'll cite a part 
of it at the end of the message. As of the supervision, every higher level 
supervises the lower, the consciousness is just on the top. [If you mean not 
just subconscious, but sub-cortical - well, they work all the time anyway, and 
they are quite rational, just their cognitive capacity is much smaller. Fight 
or flight is rational (goal-directed), reaction to pain is also very 
reasonable, the conditioned fears as of the "Little Albert experiment" and 
other phobias are also rational and right for themselves from the perspective 
in the time when they were created, i.e. if its known how those "irrational" 
phobias have started, they stop being "irrational". Sometimes they are unknown 
or untraceable by the observer or the person but that doesn't mean 
non-algorithmic or non-goal-directed.


You play the keyboard, aren't you(?).
Hawkins have made this hypothesis in "On Intelligence", and I have done it 
since to me it's an obvious observation, before his book in a publication, that 
practice/mastering of particular domain moves it from the higher cognitive 
levels (and consciousness) down to the lower, i.e. it gets less "conscious", 
i.e. it goes into the *crystallized* intelligence. I think that's well proven 
fact in experiments with chess players' and other experts' vs non-experts' PFC 
patterns - if I recall correctly, the brain of one with experience uses a 
"bank" of cases and doesn't fire much, while the one without experience have to 
search from scratch each time and her brain "explodes" - it doesn't know what 
to do, and this is supposed to be "conscious", i.e. exploration, while the 
expert "just know", i.e. he moves "intuitively". Saying that it's 
"subconscious" or "irrational" is a nonsense, that experience or its precursors 
have been conscious and goal-oriented in the past, then it's processed and 
stored for fast access.



Other well known facts are the evolution of the skills in riding a bike, 
driving a car, juggling or playing musical instruments.
The better improviser you get with a guitar or piano, the more unconscious you 
become while improvising, it becomes more and more "automatic" - however that 
doesn't mean "uncreative" for non-qualified observers, it's still creative, 
improvisation is stamped as "creative", but the top-level parts are too slow to 
cope with it, they see it with a delay.


The most untalented player needs the most of concentration and he is the 
slowest, because he has to rethink every tiny move with his most slow and 
clumsy circuits. The same goes for playing computer games - the best playing, 
especially in FPS games, often comes when you *stop* thinking and let your 
"subconscious" work, i.e. after you leave *faster* circuits to deal with the 
problems, which are simple.


That is, higher intelligence and higher talent activities make humans look more 
"unconscious" and less of control of their actions, you also can't think of 
them in real time - it just gets too fast for consciousness to think about it.


And that was how I concluded, that consciousness is apparently not required 
neither for intelligence, nor for creativity, which in my theory are a part of 
the same procedures. There are other examples, too.



Mike> "Creativity is a whole different culture to that of rationality."


It's not, it's the same, if you do both, like me. The answer to this is even in 
my works from 2002-2003-2004, but the ignorance of others is widespread. 


Science = Art + Fun 
;)http://3.bp.blogspot.com/_yeqjAlu3lyQ/SmIjetZgJII/AAAAAAAAAbY/cKQEDUNy3uE/s1600/noshtni_oblojka_7_900.jpg



Recently I found a paper, which presents my decade older theses and claims 
about the same process going in science in art in a "scientific" packing, I'll 
probably pack my response it in a paper, too.



P.S. Regarding "rationality" and the "irrational subconsciousness" and the 
psychiaters (who apparently have missed the classes in neurobiology, which in 
the time of Freud just didn't exist),  an excerpt from the publication at: 



Nature or Nurture: Socialization, Social Pressure, Reinforcement Learning, 
Reward Systems: Current Virtual Self - No Intrinsic Integral Self, but an 
Integral of Infinitesimal Local Selfs - Irrational Intentional Actions Are 
Impossible- Akrasia is Confused - Hypothesis about Socialization and 
Eye-Contact as an Oxytocine Source

http://artificial-mind.blogspot.com/2012/11/nature-or-nurture-socialization-social.html


 "Akrasia" as doing something "against own good will" is Confused


As of the "akrasia", I'd partially challenge the concept. IMO the philosophical 
confusion comes from the lack of physiological knowledge, wrong assumption and 
overgeneralization. There's no integral self, i.e. the brain is not an integral 
system.


It self-organizes and integrates the parts, because they are connected to each 
other, but this happens at the expense of "bugs" and apparently "irrational" 
behavior, because brain was not created at once and those integrations and 
effects were not planned.


Body and repeated sensations of self integrate "self" in the POV of the 
prefrontal cortex, and of an external observer. However there are many 
competing subsystems that are patched over each other, the highest level 
"executive function" is strongly influenced and entangled with older systems, 
which creates a mess of mechanisms and motivations. The limitation of the body 
actuators (and of the basal ganglia) reduce the possible physical actions and 
make the body appear as having an integral personality/mind/soul.


Philosophers who are searching for a global and valid-all-the-time 
non-contradicting integral "will", "moral", "good will" for all possible cases, 
face those paradoxes of "doing something against one's better judgment" (as 
cited in the Wikipedia article).


Integral of Infinitesimal Local Selfs over given Period...
Current Virtual Self - A Snapshot of the Virtual Simulators in the Brain


I've discussed in (see... 2002, 2003, 2004, Analysis of the sense... ) that if 
you do something intentionally, that means without your hands being pulled with 
a wire from another explicit causality-control unit (an agent), or without 
another agent to force you with a loaded gun etc., then that's what your 
current virtual self/"will" has chosen as the best action given the experience 
and the possibilities it understands, and given the time-span and rewards that 
it sees from its own perspective, at this very specific moment of 
decision/action, computed for a selected time-period etc. That self is virtual 
and "exists" at the moment of acting, e.g. moving your hand, grasping something 
etc. In the next moment there might be another virtual self, which has other 
goals and motivation, which are valid for the next moment, but they might be 
"inconsistent" with the past or the next, because the underlying model is 
covered under the skull and in the long history of experience.


An analogy can be an Integral of Infinitesimal Local Selfs, in Calculus terms - 
a Calculus of Self...
Sometimes, for some cases, in some situations, different virtual current selfs 
match and are/appear as stable, because the set of possible actions is limited, 
and because brain has also stable parts and configurations as well (at certain 
resolution), but the point is that "irrational" and "not-consistent" actions 
are not really such. I claimed in those papers, and still claim, that 
"irrational voluntary action" is a nonsense. 


If something seems "irrational", that means that the observer hasn't recognized 
either the correct agent, the correct "rationality" or both, or hasn't done 
with sufficient resolution in order to predict it right. The concept of 
"rational" (as "consistent") is confused and primitive.


Due to the mess in human cognitive and physical reward system *, the moral 
values can change all the time and the "good or bad" - too, especially if it's 
something "abstract", i.e. not directly linked to feeling of dopamine, 
oxytocin, etc. which can have a very fast effect.


Some philosophers don't get it and treat self as a constant, it's like 
integrating a constant - it equals 0.
Brain is not abstract and constant, it's more like a complex (complicated) 
function - it has specific needs at specific moments, which are caused by 
specific sensations stored now or before 10 years in specific circumstances 
etc. which are associated with specific physical sensations ("gut feelings", 
projected eventually to the insular cortex**). 


Brain constructs generalizations out of those specific experiences, but there's 
a lot of noise and variations, and also working and short-term memory (recent 
activities and experiences), the environment of every precise moment and the 
declarative/autobiographical memory contain many specifics, which can be called 
internally in a sequence that seems "random" for an external observer, while it 
may have it's very specific reasons, grounded in experience.


Such an observer, - who is assuming "rationality" wrongly as something that he 
believes is "good", "best" etc., rather than what's best for the agent's own 
estimate, - wrongly concludes that if somebody breaks his apparently wrong 
model, he acts against "his good will". NO, it acts against the WRONG model, 
following its own will. If an agent does something "against his will", then 
that's not his will.


"Will" is considered as something abstract and independent from the body, e.g. 
if "you want to quit smoking, but you don't", therefore "you have a weak will". 
In fact yes, it is separated from the body as the decisions may be initiated by 
the PFC, and the statements of will might be just words, while the real 
non-verbal actions are driven by lower dopamine-shortcuts, such as nicotine 
addictions.



* We've discussed this on the AGI List, see also below ** See also Damasio's 
works

Akrasia, as "watching too much TV, realizing that it's a waste of time" or 
"eating too much and not practising sports, knowing that it causes obesity" - 
in my opinion there are simple reasons and I don't think the reasons have been 
much different in the past.


Do the average people 100 years ago used to study Vector Calculus, Maxwell's 
Equations or did they constructed cathode-ray tubes or radio equipment or did 
they studied all kinds of sciences in order to make new inventions, instead of 
just going to the pub, theater, cinema, chatting, flirting, reading newspaper 
articles about crimes and random news from the world?


The reason why they didn't and why they preferred simpler "social" activities, 
is that the intellectual activities require cognitive profile and capacities 
that a small minority of the population have, and the long-term goals are hard 
for the mind even for the gifted ones. One reason - the relation to present or 
to the future present is questionable and unclear - as noted in the famous 
Einstein's quote about people that love chopping woods, because they can see 
the results right away.


In physiological terms, there are dopamine shortcuts, or we may call them short 
circuits - humans *are* "wirehead"s - which are making long term activities 
harder, at the expense of short term ones. 


There are easier, simpler and cheaper short-term activities providing the 
desired "drugs", why shooting for something long-term that's uncertain?
The long term ones have to have some kind of immediate measurable effect in 
order to keep the interest and compete with the activities which provide 
feedback immediately. There we are some of the effects of the clumsy AI/NLP and 
other fields in the academia, where small, incremental and "completely provable 
immediately right now, with no delay" results must be presented, even if they 
are globally very vague or meaningless.


It's also an illustration about how bad and weak human brain's executive 
function is, and how pathetic working memory can be - that's one reason why we 
need to take notes and pin to-do notes on the wall.


(...)

-- 


.... Todor "Tosh" Arnaudov ....


.... Twenkid Research:  http://research.twenkid.com



.... Self-Improving General Intelligence Conference: 
http://artificial-mind.blogspot.com/2012/07/news-sigi-2012-1-first-sigi-agi.html


.... Todor Arnaudov's Researches Blog: http://artificial-mind.blogspot.com






  
    
      
      AGI | Archives

 | Modify
 Your Subscription


      
    
  







  
    
      
      AGI | Archives

 | Modify
 Your Subscription


      
    
  

                                          


-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657
Powered by Listbox: http://www.listbox.com

Reply via email to