• Choose a category
• All
• Classic-Fiction

Physics and neuroscience both tell us, for different reasons, that one clump of matter can’t be about another clump of matter. Computer science combines both to show that human brain states can’t really be about stuff for exactly the same reason that the internal workings of your laptop can’t really be about anything at all.

Suppose you are reading this book on a Kindle or an iPad, or even on an old-fashioned computer screen. On the screen there are pixels spelling out the sentence

NEVER LET YOUR CONSCIOUS BE YOUR GUIDE.

Why is this arrangement of dark and light squares on the screen—this clump of matter—about consciousness, or about guiding you, or about never letting the former be the latter, or about anything at all? Obviously, because you have interpreted the pixels as being about consciousness and about guiding and about never letting something happen. Computer hardware designers and computer software engineers designed the pixel configurations to represent letters and words and sentences—that is, to be about letters and words and sentences. You look at the pattern of pixels and interpret it to be about the subject of the sentence formed by the pixels.

Going back even further in the design process, electrical engineers interpreted the CPU’s (central processing unit’s) semiconductor circuits as “logic gates”—that is, as being about “and” and “or” and “not.” Without interpretation by someone, neither the pixels nor the electrical charges in the computer’s motherboard nor the distribution of magnetic charges on the hard drive can be about anything, right? They are just like red octagons. They get interpreted by us as letters spelling English words. That’s how they come to be about the things that the pixels are about. So, no aboutness anywhere in the computer without interpretation by us—by our mind, our brain.

When a computer gets programmed, the only changes in it are physical ones: mainly the distribution of high and low currents or north and south magnetic charges to its circuits or magnetic media. These changes in voltage or magnetic pole store both programs and data. Together, they determine output, usually the pixels on the screen.

After building a computer that beat the world chess champion several times, the computer scientists at IBM began seeking a greater challenge. They sought to program a computer, named Watson (after the founder of the company), to play the American quiz show Jeopardy!. They built a computer that not only stored as much (or more) information as a human brain, but could deploy that information appropriately—that is, produce pixels that humans interpret as questions to Jeopardy! answers. The achievement was far more impressive than merely beating Boris Spassky. But the computer scientists were just arranging Watson’s circuits to respond to voltage-pattern inputs with voltage-pattern outputs. When they scanned vast quantities of data into its memory circuits, they were endowing its circuits with information, but not with thoughts (or anything else) that are about stuff.

It’s remarkable that Watson is pretty good at Jeopardy!. Yet its electronic circuits don’t need to be about anything, including about how to play Jeopardy, for it to be pretty good at the game, good enough to beat most people.

How does Watson manage to do it without having any thoughts about stuff? How does it search through its data for the right response when its data consist of nothing but vast assemblages of high- and low-voltage currents in microchips and distributions of magnetic charges on hard drives? How does it even understand the quizmaster’s question—or rather the answer, since this is Jeopardy!—without having thoughts about anything at all, without even thoughts about the language the question is asked in? Computer scientists know how Watson does it. How vast is the information stored? Roughly as vast as an educated adult human stores for Jeopardy!-type uses. How intelligently does Watson deploy it? Approximately as intelligently as