\documentclass{slides}
\usepackage{color}
% usage:  \textcolor{cyan}{ Reasoning about action}
% for 3400 display, compile on Powerbook 3400
% configs on Oztex, select A4 landscape
\setlength{\textwidth}{8in}
\setlength{\textheight}{6in}
\setlength{\topmargin}{0in}
% colors include cyan, red, green, yellow (pale) magenta
\newcommand{\commentout}[1]{}
\begin{document}

\begin{slide}

\begin
{center}
        \textcolor{red}{THE WELL-DESIGNED CHILD}\\
        John McCarthy\\ Stanford
        University\\ jmc@cs.stanford.edu
        \\\textcolor{red}{http://www-formal.stanford.edu/jmc/} \\ \today\\
        \vspace{0.3in}
\end{center}
% these slides started 2007 Sept 21 for Rome talk

$\bullet$ The world into which a human is born is complex.  We'll
discuss some of the complexities, concentrating on one---3-d
semi-permanent, movable objects.

$\bullet$ Evolution has equipped a baby with some innate knowledge of
the world.  That works better than starting as a blank slate.

$\bullet$ Consider a well-designed logical robot child (WDC).  Mostly
it will have the innate abilities we conjecture real children have,
but sometimes we can do better.

%My papers are on the above web page.  This paper is \\
%\textcolor{red}{http://www-formal.stanford.edu/child.html}.

%\textcolor{blue}{
%``the well-designed child'' is from Stephen Pinker's \emph{The Blank
%  Slate}.  If I had it handy, I'd include a quote.}
\vfill
\end{slide}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


\begin{slide}

\begin
{center}
        \textcolor{red}{THE LOCKEAN BABY}\\
        \vspace{0.3in}
\end{center}

$\bullet$ John Locke 1693: The baby starts out as a blank slate.  It
builds its knowledge by inferring associations among stimuli.

$\bullet$ Since 1950 people proposed to start with a Lockean blank
slate baby machine and have it learn from its experience.

$\bullet$ Starting from the blank slate, I'll bet it's a lengthy
process to infer the existence of 3-d objects.  I don't think anyone
has made an AI system that can do it.

$\bullet$ A billion years of evolution has provided us with pretty
good prejudices about the world far better than a blank slate.  Our
robot child should also have them

\vfill
\end{slide}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


\begin{slide}
%\vspace{-0.5in}
\begin
{center}
        \textcolor{red}{THE WORLD IS COMPLICATED}\\
        \vspace{0.3in}
\end{center}

$\bullet$ The world's structure is not directly describable in terms
of the input-output relations of a person.  The basic structure of the
world involves elementary particles on time scales of $10^{-25}$
seconds, but intelligence only evolved in structures of more than
$10^{23}$ elementary particles.  

$\bullet$ Even at the level at which a small child can  perceive it,
the world is  extremely complicated.  Here are some of the complications.

$\bullet$ \textcolor{blue}{Reality and appearance} Animals and humans
don't perceive the structure of our environment directly.  Senses have
evolved to give \emph{partial} information about objects and their relations.

$\bullet$ The world is 3-d, but our senses react to surfaces.

\vfill
\end{slide}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\begin{slide}

\begin
{center}
        \textcolor{red}{MORE COMPLICATIONS}\\
        \vspace{0.3in}
\end{center}

$\bullet$ \textcolor{blue}{semi-permanent objects} Much of the world
consists of three-dimensional objects that have masses, moments,
compliances, hardnesses, chemical composition, shapes, outer surfaces
with textures and colors, are often made of identifiable parts which
sometimes move relative to each other.  A particular object can
disappear from perception and reappear again.  

Note that the structure and location of an object in the world are
more persistent than its appearance and location in the visual field
or relative to the hands.

\newpage $\bullet$ \textcolor{blue}{temporal structure} The
environment of a child has a complex temporal structure.  Some items
change in seconds, others last for hours, days, or years.  Babies live
in the present.  The concepts of \emph{tomorrow} and \emph{yesterday}
are not learned right away.

$\bullet$ \textcolor{blue}{causality} Events cause changes in objects
and their relations and cause other events.

\vfill
\end{slide}

\begin{slide}

\begin
{center}
        \textcolor{red}{STILL MORE COMPLICATIONS}\\
        \vspace{0.3in}
\end{center}


$\bullet$ \textcolor{blue}{solidity} Objects are solid and do not
ordinarily penetrate one another.  Some are rigid and some are
flexible.
  
  $\bullet$ \textcolor{blue}{gravity} Unsupported objects fall to a
  lower surface.
  
  $\bullet$ \textcolor{blue} {kinds of objects} Objects have kinds,
  and objects of the same kind have properties associated with the
  kind.  Babies are ready very early to learn what kinds there are.
  
  $\bullet$ \textcolor{blue}{natural kinds} Many of the objects a
  child encounters, e.g.  lemons, belong to {\em natural kinds}.  The
  objects of a natural kind have yet undiscovered properties in
  common.  Therefore, a natural kind is not definable by an
  \emph{if-and-only-if} sentence formulated in terms of observables.
  
  $\bullet$ \textcolor{blue}{relations} Objects not only have
  individual properties and belong to kinds, but objects and kinds
  have relations with one another.  At least some ternary relations
  such as betweenness are basic.  Also ``A is to B as C is to D''
  seems to be basic.  In its numerical use, it reduces to the equality
  of two fractions, but the quaternary relation seems to be basic in
  common sense usage.

\vfill
\end{slide}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\begin{slide}

\begin
{center}
        \textcolor{red}{WHAT DO BABIES KNOW AND WHEN?}\\
        \vspace{0.3in}
\end{center}

$\bullet$ \textcolor{red}{Q: If the world is so complicated, how can
  babies do anything purposeful?}

$\bullet$ \textcolor{blue}{A: They know simple cases of phenomena.}

$\bullet$ There's good psychological evidence that they have an innate
knowledge of solid objects that continue existing even when out of
sight.  \textcolor{blue}{How do they remember an object that has
  gone out of sight so as to recognize it when it reappears?}

$\bullet$ They are ready to ascribe purposeful action to humans and
animals and to try to influence them.

$\bullet$ \textcolor{blue}{the principle of mediocrity} (from the
astronomers) Our child is like other children.  This lets it reason in
both directions.

$\bullet$ \textcolor{blue}{persons} Some objects are animate and have
purposes analogous to those of the child.  They can be influenced but
are sometimes to be feared.  Defects in innate understanding of
persons, e.g. autism, harm the child.
\vfill
\end{slide}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\begin{slide}
\begin
{center}
        \textcolor{red}{WHAT DO WE WANT IN A WELL-DESIGNED ROBOT CHILD?}\\
        \vspace{0.3in}
\end{center}

$\bullet$ \textcolor{blue}{Distinguish appearance from reality}

$\bullet$ \textcolor{blue}{natural kinds} Natural kinds don't have
if-and-only-if definitions.  There may always be more properties to be
learned.  To a small child, all kinds are natural.  The robot child
should think in terms of natural kinds.

$\bullet$ \textcolor{blue}{three-dimensional objects} These are more
stable than the perception of them by any sense.

$\bullet$ \textcolor{blue}{perceive motion as continuous}

$\bullet$ \textcolor{blue}{actions and their effects}

$\bullet$ \textcolor{blue}{recognize parts} Recognize parts of an
object and their relations to the others.  It would be interesting to
have a grammar of 3-d physical structure analogous to that of
sentences.

$\bullet$ \textcolor{blue}{focussed curiosity}

$\bullet$ \textcolor{blue}{grammar of goal regression} To do A, I need
to do B first, and to do B, I need to do C first.

$\bullet$ \textcolor{blue}{introspection} Children begin to do this by
age 3, do it well by age 5.  The WDC needs it.

\vfill
\end{slide}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\begin{slide}

\begin
{center}
        \textcolor{red}{THE SPELKE EXPERIMENT}\\
        \vspace{0.3in}
\end{center}

This psychological experiment exhibits abilities of human babies we
want in the WDC.

Elizabeth Spelke described a number of experiments that she and others
did to discover and verify innate mental abilities.  The technique
uses the fact that a baby will look longer at something surprising
than at something that seems familiar.

Here's one that was first done in 1973 and was repeated by Spelke in
1993 with two months old babies.  There are experimental babies and
control babies and the experiment has two phases.  In the first phase
the control babies are shown nothing.  The experimental babies see an
object go behind a screen and shortly another object emerges on the
other side of the screen.  The timing is such as would be appropriate
if the first object struck the second object and knocked it from
behind the screen.  The babies are shown the phenomenon enough times
to get bored with it and stop paying attention.

In the second phase of the experiment the screen is removed.  There
are two variants.  In the first variant, the first object strikes the
second and knocks it onward.  In the second variant the first object
stops short of the second, but the second object takes off as though
it had been struck.  The control babies look at both variants for the
same amount of time.  The experimental babies look longer at the
second variant.

The conclusion is that the experimental babies inferred that the first
object had struck the second when the event occurred behind the
screen.  When the screen was removed, they were not surprised when the
expected event was shown to occur but were surprised and looked longer
when this expectation was not met.

The conclusion is that babies have innate expectations about dynamics.
For details see Spelke's 1994 article in \emph{Cognition},
\emph{Initial knowlege: six suggestions}.

\vfill
\end{slide}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\begin{slide}

\begin
{center}
        \textcolor{red}{THE WELL-DESIGNED CHILD AS A LOGICAL ROBOT}\\
        \vspace{0.3in}
\end{center}

$\bullet$ Not even a sketch of a design---just some ideas.

$\bullet$ Appearance and reality

$Appears(appearance, object)$ is too simple except in a limited
context, but children think in limited contexts.\\
%There isn't time for the logic of contexts in this lecture.
$Holds(Appears(person,appearance, object),s)$ says more and is
suitable for referring to the child from the outside.  

\vfill
\end{slide}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\begin{slide}

\begin
{center}
        \textcolor{red}{CONTEXTS}\\
        \vspace{0.3in}
\end{center}

$\bullet$ The correctness of a child's beliefs and references to
objects depends on context.  Children change context frequently.

$\bullet$ When looking at a child's ideas from the outside, we can use
the theory of contexts as objects introduced in my \emph{Notes on
  formalizing context}.  In that theory $Ist(c,p)$ is true if the
proposition $p$ holds in the context $c$.  $Value(c,exp)$ is the value
of the expression $exp$, e.g. $Color(Block1)$, in context $c$.

$\bullet$ $Value(Beliefs(Child1,S0),Color(Block1)) = Red$ asserts that
in the context of Child1's beliefs in situation $s$, Block1 is red.
The context theory lets us \emph{enter} the context
$Beliefs(Child1,S0)$.  Then we have $Color(Block1)=Red$.

\vfill
\end{slide}
% $Value(Of(x),f) = f(x)$
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\begin{slide}

\begin
{center}
        \textcolor{red}{THREE CHALLENGES AND A CONCLUSION}\\
        \vspace{0.3in}
\end{center}

$\bullet$ Make a robot baby that can do the Spelke experiment
honestly.  But what's honestly?  It may start knowing about collisions
and knowing about occluded objects, but at least it shouldn't start
with knowledge combining the two.

$\bullet$ Its contexts should change like those of a baby.

$\bullet$ It should properly relate 2-d appearances to 3-d reality and
also relate tactile appearances to reality.  Here's a puzzle.

$\bullet$ \textcolor{blue}{What babies know presents challenges to
  AI---to AI of all schools.}

$\bullet$ This talk is partly based on the manuscript\\
\textcolor{red}{http://www-formal.stanford.edu/jmc/child.html.}

\vfill
\end{slide}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\end{document}



\begin{slide}

\begin
{center}
        \textcolor{red}{}\\
        \vspace{0.3in}
\end{center}
\vfill
\end{slide}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Cartesian baby, designer stance

slides
blank slate vs. realism

the world
the innards of objects

relation of the senses to the facts  - perception