The first order of business is to instantiate TTC. To do so, we will make
use of Schacter's [11] (partial) model of the mind, which
is summarized in Figure 2. In order to clarify this model, we find it
helpful to recast it as a model at home in the field of
Artificial Intelligence; the recasting is shown in Figure 3.
Effectors are those parts of the system that allow for it
to do such things as
move around its environment (e.g., feet, or wheels in the case of a
robot); sensors are organs (e.g., eyes
and ears) devoted to bringing in information from the
environment.
We follow Nagel [8]
in roughly characterizing states
of phenomenal consciousness as those which are such that there is something
``it is like" to be in them.
Schacter's model, as desired, is an ideal representative of nearly everything Dennett opposes in these matters. (Notice that Schacter's model appears to commit to the Cartesian Theater: the place where the executive controller does its work.) Whereas his MDT would have consciousness distributed throughout the system in haphazard fashion, Schacter's model has this key property:
Let's move now to instantiate E. Note, first, that if the spots of color in
the color phi experiment are separated by an interval of time of sufficient
length, no apparent motion results: in this case the spots of color appear
to the subject to be mere motionless flashes, first read and then green.
Now, our instantiation consists of
a Turing Machine (TM) specification of the dotted enclosure shown in Figure 3.
More specifically, we provide a TM-based sub-system that, playing
the role of the phenomenal consciousness module, operates
over the information coming in through the eyes (in the case of a human
experiencing color phi) to produce output which the executive system
``sees" as indicating motion from red to green, or -- if the
delay between the appearance of the spots is long enough -- mere flashes of red and
green. Which is ``seen" depends on the interval of time between the appearance
of the spots. The overall architecture of our TM model is shown in
Figure 4. The basic idea is remarkably simple -- so simple
that we have implemented the model in the form of an actual (composite)
Turing Machine in the well-known Turing's World 
software.
This machine, 
, is available from --'s
web site (for the URL see note n); it is shown
in flow graph form in Figure 5.
That which appears on the
machine's tape at the start of processing corresponds to what the sensors (eyes)
detect in this case. For example, the string
represents the flash of a red spot for 150msec, 50 msec during which nothing is flashed (the blank space between R and G), and then the flash of a green spot for 150msec. (The symbols # flanking the string simply indicates the endpoints of that which the subject's eyes detect; compare the schematic here with the snapshot shown in Figure 1.) If the machine takes in this string, it returns the (metaphorical) string
for the executive controller. On the other hand, if the machine takes in the string
it will yield for the executive controller
which indicates mere flashes without apparent motion.
That is consistent follows by inspection. This means that our
instantiation of E is consistent. We are assuming, unexceptionably,
that our
Schacter-based instantiation of TTC is consistent. That which the
executive controller receives from corresponds to that which
the person in the color phi experiment ``sees," so we have established
It follows by our schema that Dennett is wrong: TTC is consistent with the color phi phenomenon.
