Strategy and Agenda:
The focus is first on the “easy” problems, with the hope that coping with them may help
(through developed ideas and skills) in addressing later the more difficult problems.
Such an outcome is a “hope” more than an expectation, certainly not a guarantee.
Nonetheless, I expect to proceed first with an examination of object manipulation, then
language production, followed by cognitive development and then social development.
This is my estimate of the relative difficulty of the problems, more than an estimate of the
probability of success in the initiatives. About sensory motor development, as essential
as its study may be, I have little hope for new insights from this corpus. Piaget had many
ideas, which we expect to follow, but with little commitment or elaboration.
Variable Depth Transcription and Analysis
The basic challenge is answering how we can get sufficient transcription for good access,
either by textual search or by what the behavior means — without crushing the effort with
excessive and unneeded detail. Here’s how we begin, and the transcriptions of P146 exhibit
an example — with three components: the news, the detail, and the meaning.
The “News” indicates the “who, what, how, where, when”… and possibly the “why,” if obvious.
A playwright might see this as “script” like description of the action transcribed.
The “Detail” goes into more minute descriptions, and this is where one would hope to
respond to Selfridge’s challenge to “describe the world so learning is easy.” [I do not have
such an example on line. Expect it soon. 10/09/13].
The “Meaning” begins to rise from the nitty-gritty of behavior to the processes witnessed;
documenting changes in those processes in their context is “the holy grail” of this research.
The immediate focus is adding “Dragon Naturally Speaking” technology to the interpretation effort.
After training the dragon system, I attempted DNS transcription of audio exported from video; this
is infeasible because Peggy’s speech was both un-comprehended and interfered with transcription
of adult speech. Consequently, even though a little strange, I will proceed with “echo Transcription,”
playing the video on one machine (listening with earphones) and repeating the dialogue and describing
the activities of the parties for recording and transcription by DNS on the second computer. Underway
is preparation of work using this method with P146 and the Example Post, “Pure Point.”
Early Ideas and Goals:
LC3ca1 CASE: a Case Study Analysis Support Environment
LC3ca2 “Soft” Sciences & “Hard” Problems
LC3cA3 Notions from research vectors: from a Talk At BBN (10/20/09)
The 50 minute talk was largely focused on samples of the IPS video corpus.
The “notions” below, never discussed, were handouts as a sketch of the framework of my ideas.
They were developed over many years from reflections on the IPS corpus and Minsky’s Society
of Mind, most fully so in my year as Visiting Fellow and Professor at the Archives Piaget and
University of Geneva.
LC3cA31 Talk Announcement, 1p.
LC3cA32 Analyzing Learning in a Micro Culture, 2pp.
LC3cA33 An Example of Learning: culdah/VaeVae, 1p.
LC3cA34 A Language Learning Theory, 2pp.
LC3cA35 The Genesis of Symbolic Thought, 4pp.
LC3cA6X One Selfridge Model: COUNT: the interplay of path specification and chance.
Current and Future Research Vectors:
LC3cA4: Analyses: Sensory Motor Development NAY
LC3cA5: Analyses: Social Development NAY
LC3cA6: Analyses: Object & Spatial Knowledge: Development Threads (current focus)
LC3cA60: Summary Table of Standard Objects and their Uses
LC3cA61: Knowledge of Objects: Backwards Events Analysis
LC3cA61a: Analyzing Interactions (see links below): Describing Objects and Actions
LC3cA61b: Analyzing Interactions (a stub for a goal): Inferring Goals from Affordances and Outcomes
LC3cA62: Knowledge of Objects (a stub for a goal): Forward Model Construction
LC3cA63: Episodes from 36 down to 18 months: Mastery of the Ring Tower Puzzle: the issue of order
LC3cA64: Starting: at 36 months: Nesting Boxes: inclusion within concave objects; towers of convex objects
LC3cA65: After: (NAY) Nesting Cups: inclusion within concave objects; towers of convex objects
LC3cA67: Later: (NAY)Blocks, Balls, Rods & Cones: disentangling shape and phenomena of rolling
LC3cA68: Finally: (NAY)Blocks and Building: shapes and support relations
LC3cA7: Analyses: Language Development NAY
LC3cA7?: Theme: Communication and the Beginnings of Speech:
LC3cA7?: Theme: Stories for Toys:
LC3cA7?: Theme: Names, Letters, and Words
LC3cA7?: Theme: And Now Computers Too
LC3cA8: Analyses: Cognitive Development NAY
LC3cA80: Theme: Integrative Review of Ideas Proposed Previously
LC3cA9 Analyses: Objectives & Methods in Earlier Research Proposals
LC3cA91 Theme: reference to proposals of “historical interest” in LC3c Analyses NAY
LC3cA92 a Proposal to the Spenser Foundation in 2000 (not funded). Three Years and Talking
LC3cA93: NSF proposal for this work from the MIT Media Lab & Archives Piaget (not funded).
LC3cA94: Summary of key ideas from four mentors (See videos at LC0aV Five Pioneers Talk about Learning, AI,and Education).
LC3c Analyses: Vignettes by Categories: the need for separate category-based listings is obviated by the tag-intersection facility installed in the left sidebar;
with the range of temporal tags available from years (1-8), quarters (1-31), or weeks (1-255), there is the finer grain of specification that our analyses require.
Typically, one would select first a temporal tag then intersect with a category tag as below:
language development: use tag Lng; social development: use tag Soc; sensory-motor development: use tag SM;
object & spatial knowledge development: use tag Obj; cognitive development: use tag Cog; reading development: use tag Rdg;
computer use: use tag Cmp; about counting: use tag Nbr; other observations: use tag Otr.