Archive

Archive for May, 2014

Weekly QuEST Discussion Topics and News, 30 May

1.) Plan for where Lizzy is headed with her summer QuEST work – comments / advice welcome:
a. adopt personal position on events / situations / qualia – that you can present / defend to your colleagues in Aug and those attendees to the open house
b. formulate ideas associated with an ‘event’ based content management system – applicable to both our intel analysts and clinicians – (this formulation will assume the ECMS {event-based content based management system} exist – that is a means to capture events and associate all the observables associated with them and link them together and in a temporal sequence – a narrative – then formulate a means to search for similar events in the data base and envision how they can be presented / used by the human operator) —- basically step two is to envision the end state where we posit the world will be if we had such a thing as an ECMS – using the definitions from step 1
c. explain the idea of the link game – demonstrate to the uninitiated non-QuEST person – demonstrate its utility in a memory contest sort of demo (related to steps above because we think events are the chunks formed by humans via these links – we associate disparate sensory data into events – we do so via linking – so this step is to explain the use of links by humans)
d. explain and demonstrate the tie between links and events (see parenthetical statement above) – I would suspect we would define events as a linked set of observables / links also to prior experiences / events – this is related to chunking – in areas to be discussed (intel analysts / clinicians) demonstrate the chunking together of observables into events and how interaction at the event level is used
e. in applications to be determined (clinical / intel analyst) architect some examples of events with the observables and the types of links necessary to capture a useful event representation for storage
2.) Capt Amerika has spent more time this week attempting to maturing a position on the ‘unexpected query’ – the idea that current approaches to AI (recall the reviewer’s comments to Sandy’s paper on narratives and consciousness) don’t provide the means to respond appropriately to stimuli that aren’t represented in the data used in their development (data driven AI solved but what happens when I don’t have the data?) – so we will have a discussion on what is an unexpected query AND then discuss two articles
a. the first is on work related to last week’s talk on embodied and grounded cognition specifically we want to set up independent investigations by discussing the introduction to the special topic discussed in Frontiers in psychology (Anna Borghi article Aug 2011, vol 2 article 187) – the relation of this article to the unexpected query issue is one of the themes of unresolved issues that must be answered by embodied and grounded cognition people is the idea of abstract concepts
b. the second related topic is on work in ‘conceptual combination’ – Conceptual combination research investigates the processes involved in creating new meaning from old referents. It is therefore essential that embodied theories of cognition are able to explain this constructive ability and predict the resultant behavior. Article by Lynott – ‘Embodied conceptual combination – Frontiers in psychology nov 2010 vol 1 article 212
i. …, by failing to take an embodied or grounded view of the conceptual system, existing theories of conceptual combination cannot account for the role of perceptual, motor, and affective information in conceptual combination.
ii. In the present paper, we propose the embodied conceptual combination (ECCo) model to address this oversight.
iii. In ECCo, conceptual combination is the result of the interaction of the linguistic and simulation systems, such that linguistic distributional information guides or facilitates the combination process, but the new concept is fundamentally a situated, simulated entity.

QuEST Discussion Topics and News 30 May

Advertisements

Weekly QuEST Discussion Topics and News, 23 May

QuEST for 23 May 2014
1.) We want to discuss an article Words as Tools – by Anna Borghi et al. – this paper follow the ideas we’ve advocated on the mind is embodied and the body is embedded – we got this phrase from Gerald Edelman (we’ve edited it to be ‘mind’ versus ‘brain’) – but the idea is that cognition is based on the ability of the critter to interact with the world that it is embedded in – one of the issues embodied cognition zealots have struggled to answer is how do ‘abstract’ concepts get grounded – the grounding of language is discussed and then extended to include social embodiment – words can have a concrete referent label that are ‘attached’/grounded to sensorimotor individual experience but in this work doing a similar ‘attaching’ associate with the use of the social word as a tool is advocated.
WeeklyQuESTDiscussionTopicsandNews23May

WeeklyQuESTDiscussionTopicsandNews23May

Weekly QuEST Discussion Topics and News, 16 May

QuEST topics for 16 May 2014:
First note this meeting will be in Building 600 Sensors Directorate and the call-in information is different.
AFRL/RY DarkStar – WPAFB Area B – Bldg 600 – 2nd Floor – Room 2BU427
Dial in # – 937-938-2201 or 937-938-2187 (COMM)
798-2201 or 798-2187 (DSN)
Conference ID – 8525#
Passcode – 486315#
The topics include:
1.) An Article supplied to us by our colleague Kirk W. – Toward a Computational Model of Narrative – George Lakoff, Srini Narayanan, we will use this article to further our discussion of the need for our artificially conscious QuEST agents to generate a cohesive narrative to represent their stream of consciousness – and specifically focus on this work since it discusses in some detail the ideas associated with their KARMA model and its ability to capture fine grained to capture the wide range of possible events and their interactions – Note specifically the capturing of EVENTS – we have converged recently on the utility of event based representation in our PCPADx effort. Their particular approach provides us a strawman of axes that will have to be implemented in our QuEST agents to be able to facilitate processing of the unexpected query – I specifically want to focus on their use of metaphorical reasoning and event representations.
2.) The second topic is a return to ideas associated with Type 1 processing that we began a discussion of last week – the work of Ben Libet …. Specifically we need to set up a discussion of a paper by Soon et al – in nature neuroscience (provided to us by our colleague robert Patterson) – ‘Unconscious determinants of free decisions in the human brain’ – vol 11, no 5, May 2008. Abstract: There has been a long controversy as to whether subjectively ‘free’ decisions are determined by brain activity ahead of time. We found that the outcome of a decision can be encoded in brain activity of prefrontal and parietal cortex up to 10 s before it enters awareness. This delay presumably reflects the operation of a network of high-level control areas that begin to prepare an upcoming decision long before it enters awareness.
3.) The third topic is associated with a recent article “making fingers and words count in a cognitive robot” by La Cruz et al. frontiers in behavioral neuroscience Feb 2014 (again provided to us by our colleague Robert Patterson). Although the article is focused on the acquisition of numerical skills I found the approach (using cognitive models – connectionist models – to investigate the impact of multi-modal information on abstraction and generation of ‘meaning’ potentially applicable to other areas where we’ve struggled to define meaning and how it is generated.
4.) We also want to work our way towards a discussion of a recently referenced article on Larry Barsalou’s LASS – Language and simulation in conceptual processing – Theories of cognition often assume that a single type of representation underlies knowledge. Traditionally, most theories have assumed that amodal symbols provide uniform knowledge representation (e .. g .. , Collins and Loftus 1975; Fodor 1975; Newell and Simon 1972; Pylyshyn 1984) .. More recently, theories have adoptedstatistical representations (e .. g., McClelland et al .. 1986; O’Reilly and Munakata, 2000; Rumelhartet al.. 1986) Most recently, theories have proposed that knowledge is grounded in modal simulations, embodiments, and situations (e .. g .. , Allport 1985; Barsalou 1999, 2008a; Damasio 1989; Glenberg 1997; Martin 2001, 2007; Thompson-Schill 2003), while other theories have proposed that knowledge is grounded in linguistic context-vectors (e .. g .. , Burgess and Lund 1997; Landauer and Dumais 1997).

WeeklyQuESTDiscussionTopicsandNews16May

WeeklyQuESTDiscussionTopicsandNews16May

Weekly QuEST Discussion Topics and News, 9 May

QuEST topics for 9 May 2014

1.) One of our QuEST colleagues submitted a paper to a conference on Computational Models of Narratives – we need to know if anyone is attending the conference? Part of the brave new world of TDY controls – thanx if you are let Capt Amerika know – her paper received some mixed reviews. The comments by the reviewers I think should be discussed as we consider the generation of a new position paper on QuEST. Bottom line all three reviewers were very interested in the idea but wanted more detail specifically how our ideas relate to narratives. I would like to entertain a discussion on narratives / situations / qualia.
2.) The second topic is a return to ideas associated with Type 1 processing. Specifically a paper by Soon et al – in nature neuroscience (provided to us by our colleague robert Patterson) – ‘Unconscious determinants of free decisions in the human brain’ – vol 11, no 5, May 2008. Abstract: There has been a long controversy as to whether subjectively ‘free’ decisions are determined by brain activity ahead of time. We found that the outcome of a decision can be encoded in brain activity of prefrontal and parietal cortex up to 10 s before it enters awareness. This delay presumably reflects the operation of a network of high-level control areas that begin to prepare an upcoming decision long before it enters awareness.
3.) The third topic is associated with a recent article “making fingers and words count in a cognitive robot” by La Cruz et al. frontiers in behavioral neuroscience Feb 2014 (again provided to us by our colleague Robert Patterson). Although the article is focused on the acquisition of numerical skills I found the approach (using cognitive models – connectionist models – to investigate the impact of multi-modal information on abstraction and generation of ‘meaning’ potentially applicable to other areas where we’ve struggled to define meaning and how it is generated.

WeeklyQuESTDiscussionTopicsandNews9May

WeeklyQuESTDiscussionTopicsandNews9May

Weekly QuEST Discussion Topics and News, 2 May

The first topic this week is a quick overview of two DARPA efforts some of us had the opportunity to hear about from their PM this week – ICArUS and KRNS:

The second topic is a discussion on Dreaming and related topics in sleepwalking, sedation and hypnosis. Discussion topics include:

Why QuEST cares about understanding dreaming. – I think that the way to understand what the qualia actually in our heads are like is to see what gets made out them when we dream. Every component of a dream has to be a playback of the snippets of life experiences stored in the brain, the same ones we use to construct the model of the world which we generate in response to external stimuli while awake. We believe that the purpose of the brain doing the snippet replay is to consolidate the formation of relevant memories. Unfortunately that is done with replaying the snippets through the sensory channels. The brain must make a ‘plausible’ narrative to handle all processing in the sensory channels. That is a price we pay for reuse of subsystems. The dream itself is the narrative constructed to explain away the sequence of snippets….

Why we can’t recall our dreams. – Answer – from adam – it has to do with the lack of predictability of the dream – memories / perception / imagination is about generating sequences that could account for current experiences (or prior or potential future ones). Since the purpose of dreams (CA theory – generate potential links between recent experiences so they can more efficiently be exercised when needed) they often combine recent experiences into combinations that are not at all what one would expect to be experienced (not predictable). In fact perception is all about reliable predictability of a plausible (stable, consistent and useful) narrative of the current world. Dreams (like humor) are all about generating implausible sequences just to exercise combinations that might be useful (they often are not physically plausible but that isn’t the criterion for optimization). Since memory is a imagined past – it is very hard to imagine a sequence of ‘random’ experiences that have never occurred before…

Where do dreams come from. –

When does consciousness arise – in the womb, at birth or during early childhood? Article by Koch – this article has some discussion of dreams in it also

Why we dream. –

Role of sleep – One area of change is the synapses, the connections between neurons, which are altered as the brain receives stimuli. “What happens when you’re awake is you produce an overall strengthening of synapses,” Dr. Tononi said. “That’s good, because that’s how you learn.”

Dreams as a tuneup for the day. – Dreams are so rich and have such an authentic feeling that scientists have long assumed they must have a crucial psychological purpose. To Freud, dreaming provided a playground for the unconscious mind; to Jung, it was a stage where the psyche’s archetypes acted out primal themes. Newer theories hold that dreams help the brain to consolidate emotional memories or to work though current problems, like divorce and work frustrations. Yet what if the primary purpose of dreaming isn’t psychological at all?

• Neuroscientist Mark Solms explains how dreams may protect and distract our brains from the outside world and allow the body to rest.

• Finding answers in your dreams – When you fall asleep, you enter an alternative state of consciousness—a time when true inspiration can strike …

• 1 >> The act of dreaming is simply thinking about our usual concerns in a different state of consciousness.

• 2>> Dreams can be especially helpful for problems that require creativity or visualization to solve.

• 3>> By thinking about specific dilemmas before bed, we can increase our chances that we will dream a solution

Meanings of dreams. – What makes us dream? A new study suggests the impulse to dream during sleep may come from signals in the brainstem, not from the brain’s higher-order regions… taking an exam naked … prof equivalent version… wish fulfillment .. depends on your bias …

Boredom – implications for brain processing.

Birds and Dreaming. – One bird, in particular — the zebra finch, which researchers say has a sleep structure very much like that of people and other mammals.

The days events are incorporated into that night’s dreams. – Research suggests that much of what happens in a dream is unique to that dream. But some events from a person’s day can be incorporated into dreams in two stages. …

Vivid dreams –

Why we have bad dreams – Writing in the journal Current Directions in Psychological Science, Drs. Ross Levin and Tore Nielsen suggest that run-of-the-mill bad dreams are part of the brain’s method of processing emotions. In fact, they say, emotional regulation may be the primary function of REM sleep, the sleep stage during which most dreams occur.

Possibly add topics in sleepwalking / sleepeating / sedation / hypnosis …

WeeklyQuESTDiscussionTopicsandNews2May