Some risk involved in beginning with a leap; Mitchell’s Picture Theory
splinters through the title’s pun–a theory of pictures and theory pictured or
picture-able.  In the introduction, Mitchell calls the problem of the 21st
C. a problem of the image.  This opens onto difficulties with the
relationship between word and image, mapping and organizing fields of
representation, and discord between reading proper and spectatorship (3). 
Fumble them as we inevitably will, these and other differences might seem less
gnarled if we "adopt Michel de Certau’s terminology and call the attempt to
describe [them] a ‘heterology of representation’" (5).

This project–from theory-orientation of the first three chapters to the
applications in the remainder of the book–develops out of a concern with agency
and attempts to "open onto the image/text problematic" (7).  It also
expresses a function related to curriculum and theory (6) in an effort to engage
the following three questions: "What are pictures? What is their relationship to
language? Why does this matter?" (5).

"The Pictorial Turn"
In this chapter, Mitchell begins with Rorty’s idea of "turns" in philosophy; the
pictorial turn engages the spectator-scene relation; it involves stances on the
mass circulation of images and their relationship to text.  This iconology
is complicated by the formalizing of visual arts as a discipline, by resistance
to images in text (12), and postmodern upheavals of perspectival theories of
representation.  In a nutshell (the picture of a nutshell):

Whatever the pictorial turn is, then, it should be clear that it is not a
return to naive mimesis, copy or correspondence theories of representation, or a
renewed metaphysics of pictorial "presence": it is rather a postlinguistic,
postsemiotic rediscovery of the picture as a course, bodies, and figurality. 
It is the realization that spectatorship (the look, the gaze, the glance, the
practices of observation, surveillance, and visual pleasure) may be as deep a
problem as various forms of reading (decipherment, decoding, interpretation,
etc.) and that visual experience or "visual literacy" might not be fully
explicable on the model of textuality (16). 

This tangle (tango?) between spectatorship and reading jostles conventional
notions of reading as an undifferentiated activity (the common assumption:
everyone gets "reading").  The move to complicate or question (interrogate)
the spectator stands out from the rest of what Mitchell is doing here.  It
springs from the pictorial turn: where are the differentiated models for
reading?  Who’s running them off?  How are those (faintly)
differentiated models re-domesticated under the rubric of literacy (toss in
‘critical’ if you want)? Finally, Mitchell has a nice set six scenic inquiries
along the lines of "what might we notice?" (26).

"Metapictures"
In this essay "on pictures about pictures," (35) Mitchell works through "an
array" (57) of six+ images as a way to tease out the infinite relation between
image and text suggested by Foucault in
The Order of Things.  Mitchell calls his procedure "ekphrastic,"
which refers to "words about pictures" (38).  Self-referential images, like
Saul Steinberg’s The Spiral (1968), call attention to but
cannot exhaust the "nested" representational possibilities (42).  In a
section on "Dialectical Images," Mitchell works on the problem of "multistability"
(43)–the blinking of an image that seemingly toggles back and forth (without
dimming).  Can it be both?  The model here is the Rabbit-Duck that so
concerned Wittgenstein. Its "what am I?" is perpetually unresolved. 
Wittgenstein would give the Rabbit-Duck the possibility of both expressions at
once, thus "restoring the ‘wildness’" diminished by "psychology and by
photographic models of the psyche" (53). 

Mitchell calls Valesquz’s "Las Meninas" a meta-metapicture; the famous
painting implicates us in the infinite referentiality–an undying parallax. 
Is that you in the mirror observing the whole scene?  For Foucault,
the importance of the painting is its way of keeping open those possibilities
rather than fixing the relationships through naming.

At the end of the chapter, Mitchel considers "Talking
Metapictures" (64): This is not a pipe, The two mysteries, and Arcadia Ego. 
Such images suggest different effects; it’s worth noting the primacy of text in
pictures of reading.  Even in "This is not a pipe," the text tends to come
define the picture of the pipe rather than the other way around–"discourse has
the final say" (66).

"Beyond Comparison: Picture, Text and Method"
A method of imagetext must not become trapped in conventional comparisons. 
Mitchell proposes imagetext not as kind of reduction or collapsing of image into
text or text into image, but rather as a way to "pry them open" (100, 106). 
Furthermore, a method of imagetext understands the varied disciplinary
expectancies: these ideas might unsurprising to art historians who have long
listened to images or literary scholars for whom imagery is passe (99). 
Mitchell asks us to move beyond comparison–"the ideal trope for figuring
‘action at a distance’ between different systems" (85).  Why?

If the relation of the visible and the readable is (as Foucault thought) an
infinite one, that is, if ‘word and image’ is simply the unsatisfactory name for
an unstable dialectic that constantly shifts its location in representational
practices, breaking both pictorial and discursive frames and undermining the
assumptions that underwrite the separation of the verbal and visual disciplines,
then theoretical pictures may be mainly useful as de-disciplinary exercises
(83).

Terms: iconology (24), multistability (45), hypericon (28), ekphastic
procedure (38), gestalt (42), nesting (48), assemblages (49), figure and dead
metaphor (66), textual repetition and defacing (69), bands of readability (71), apotropaic
image (78), attributed masteries (63), calligram (77)

Effects (not to be confused with theme or topoi) (74): Pipe Effect (74), Vortex
Effect (75), Medusa Effect (78)

Figures: McLuhan (15), Panofsky (16), Foucault (5, 18, 60), Jonathan Crary
(19), Althusser (29), Wittgenstein (60), de Certau (5, 71), Deleuze (70)

"Is iconology, in contrast to its ‘disintegrative’ methodological cousin,
philology, incapable of registering ‘faults’ in culture, the fractures in
representation, and the resistance of spectators?" (23).