Jonathan Waskansecondary ones do not. As he suggests, color qualia may have more to do with what light does to
us; they may reside only in us. But primary properties are both in us and (much of the time) out
in the world. All of this is just to get to the point that there is also a mystery of how brains pro-
duce experiences of shape, motion, etc., a mystery of primary qualia alongside that of secondary
qualia. In neither case, primary or secondary, is there anything about neural activity that betrays
their instantiation. Indeed, just looking at the low level of neural circuitry, it is hard to see how
brains even could produce primary qualia (see Waskan 2011).
But this is where computer simulations of physical systems are so instructive. They offer a
framework for making sense of how electronic (or electrochemical) circuitry can give rise to
distinct properties at distinct levels of activity, from logic gates, to numerical memory registers,
to programs, to evolving coordinate specifications, to high-level models (or what have you) of
primary properties. If our high-level internal models produce experiences of primary proper-
ties in us, then we should not look outward like Noë, Lycan and Tye, or downward like Leibniz
and (as it turns out) Searle. We should be looking upward, past several levels of processing, to
internal world models. Already inspired by computers, this view lends itself well to the creation
of mechanical contrivances such as robots that have similar world models and, thereby, possess
primary qualia.
But it is not obvious at first glance how this proposal could be extended to account for, let
alone allow us to reproduce, secondary qualia. Still, it would be cosmically peculiar if primary
qualia (e.g., the experience of a tomato’s shape) turned out to be just the characteristic of some
high-level internal model while secondary qualia (e.g., the experience of a tomato’s redness)
were best explained as, say, the physical properties of quantum events or, as Jackson (1982)
hypothesized, as the non-physical exhaust of neural events. So if the above account of primary
qualia proves out, we should expect other sorts of qualia to admit of a similar, high-level solu-
tion. And we should expect to make little progress in understanding qualia of any sort as long
as we hold fast to our myopic lower-level wanderings. And again, such a high-level explanation
for secondary qualia might be easily adaptable to the creation of robotic contrivances possessing
secondary qualia like colors, smells, and love or hate.^4 Indeed, as already mentioned, the pos-
sibility of synthetic, robot consciousness may require a high-level account of qualia such as this.^5
8 How Our Robot Overlords Might Overcome Cognitive ClosureAs tempting as all of this seems to one of us, we must still allow for the possibility that the truth
about qualia is far different from anything yet proposed. Indeed, it may be so strange as to remain
effectively closed to feeble-minded creatures such as ourselves. Even so, there is another lesson
to be gleaned from the advent of sophisticated computer models of the world, one that gives us
renewed cause for optimism.
Consider that we began to produce finely detailed external models in the first place precisely
because we are so feeble minded. Even if we understand, individually, how particular bits of mat-
ter can be configured relative to one another and the laws governing their behavior, it is often
far beyond the powers of human imagination to envision in any detail how vast arrangements of
low-level events give rise to higher-level phenomena. In isolation from artifacts and technology,
plausibly our internal models of systems are always incomplete, piecemeal affairs (Norman 1983;
Keil 2006; Hegarty 2004) and what passes for unaided mechanical understanding only extends
only so far as we can find ways of compressing complicated information into useful approxima-
tions (Schwartz and Black 1996) or into simple and suggestive metaphors (Brown 2003).
However, with the aid of scale models and, more recently, computational models, we can
watch complicated scenarios play out, interact with and manipulate them, and draw inferences