^A A hard drive (HD) stores information as a series of small pieces of magnetizable materials called ferromagnetics. Under the influence of a magnetic field, these small spots (called domains) can be oriented in one of two directions (up or down). The hard drive stores a series of zeros and ones as a sequence of up and down domains. A compact disc (CD) stores information as a series of small pits in a colored plastic, making a series of dark and light spots. Hard drives and CDs can store the exact same series of zeros and ones. Although instantiated in the physical realm (magnetic domains, dark/light pits), the actual information is the sequence of zeros and ones.

^B The actual mechanism by which a cell reads the information stored in DNA is extremely complex, with new mechanisms discovered over the past several decades where the physical environment within and around the cell changes which components of the DNA are read.³ For example, specialized proteins called histones can block part of the DNA from being read, while other proteins called transcription factors can encourage part of the DNA to be read. At this point, geneticists have created entire logic control structures (if this protein is present but that protein is not, then read this part of the DNA to create a third protein). The physical environment of the cell can change which part of the program is read. But this is not that different from current computer programs, which have mechanisms to check physical aspects of the computer to decide what parts of the program need to be run. For example, when your computer starts up, it checks to see if you have a DVD reader installed or a network card. (These days, programs don’t tend to tell users what they are doing, so it may just show you a little blue bar cycling around or a corporate logo to look at, but what it’s really doing is checking for hardware.) If it finds that hardware, it installs appropriate drivers (small pieces of software) for it.

^C Even though early (1979) artificial intelligence programs outperformed typical medical faculty in controlled comparisons, these programs were abandoned due to concerns about legal issues—Who would be responsible for errors in diagnosis? A similar concern may be limiting successful automatic driving programs.⁹

^D At the time that Turing proposed his test, getting input and output from computers was very complicated, and Turing spent a large part of his original proposal ensuring that the test can’t be fooled by seeing or hearing the person or computer. We are now very familiar with the concept of communicating with invisible entities over computer networks, and, in fact, the Turing test regularly occurs in the context of videogame interactions, where some of the characters are computer-controlled and some are human-controlled.

^E Apparently, this was, in fact, how the first version of Newell and Simon’s General Problem Solver was implemented. Computers were too difficult to program at the time (in 1956), so they encoded it on 3 × 5 index cards and had their families manipulate the cards symbolically by specific rules, thus producing, literally, the Chinese room.¹⁵

^F The symbol-manipulation hypothesis is a hypothesis about how we process information.¹⁹ It is wrong—not because it cannot be true, but because it doesn’t seem to be how the brain processes information.

^G Traditionally, patients are referred to by their initials to preserve their privacy. When H.M. died in 2008, his full name was revealed.³⁶ But even before, known only as H.M., he was arguably the best-known brain lesion patient in neuroscience, rivaled only by Phineas Gage, whom we met in our discussion of frontal cortex and emotion (Chapter 8).

^H In highly interconnected tissue (such as the hippocampus and neocortex), excitatory recurrent connections (which encourage neurons to fire) are balanced by local negative recurrent connections (which discourage neurons from firing). If the excitation and inhibition are out of balance for a transient moment, you can get runaway excitation leading to massive firing of neurons.³⁸ Interestingly, seizures can arise from too much transient excitation or too much transient inhibition. Brain function generally sits in dynamic balance, which makes it easier to respond quickly and makes the brain more flexible in its responses, but it also means that a transient event takes time to return to normal. The epileptic event is a phenomenon known in the physics and electrical engineering fields as “ringing”—the system gets pushed too far to one side, and as it returns to its base point, it bounces too far to the other side, much like a spring that oscillates when pulled or pushed.

Thus epilepsy is incorrect, extraneous firing of neurons. Since the firing of neurons is memory, perception, and cognition, this extra firing of neurons will be perceived by other parts of the brain as real. Epileptic seizures in the auditory cortex are often preceded by patients hearing music, events in temporal-lobe structures can trigger memory, and seizures in emotional parts of the brain (such as the ventral frontal cortex) can trigger feelings.³⁹ Dostoevsky (a noted epileptic) described seizures in his novel The Idiot as being preceded by an emotional moment of exceptional clarity of joy, hope, and vitality.⁴⁰ This emotional feeling of deep religious connection has been seen in some modern patients with temporal-lobe epilepsy as well.⁴¹

^I These prism glasses may be familiar to some people visiting science museums. A typical neuroscience game shown at science museums and during presentations to school students is to have a volunteer toss a beanbag into a basket with prism glasses on. Because the prism shifts the visual world to the left by some angle (say 20 degrees), the person typically misses the basket by 20 degrees in the opposite direction. With continued tries, the person can learn to compensate over the course of a few minutes. And then, when the prisms are removed, there is an after-effect where the person misses the basket by 20 degrees in the shifted direction, which takes a few minutes to correct. With more experience with and without prisms, the time it takes to shift decreases.⁷¹

^J Replay in nonhippocampal subcortical structures, such as the dorsal system in the basal ganglia (Chapter 10), has not yet been reported, but there has not been the same level of extensive study done on these structures during sleep. The one exception is that the ventral striatum (the nucleus accumbens, which we saw was involved in representing potential outcomes during deliberation; see Chapter 9), replays reward-related activity in conjunction with the hippocampal replay of experienced sequences.¹⁰⁴ Sleep is also critical for procedural learning, but whether this is due to replay events in dorsal basal ganglia systems remains, as yet, unknown.¹⁰⁵

^K Replay events in the hippocampus tend to occur during an identifiable local field potential (LFP) event called a ripple or a sharp wave (sometimes called a “sharp-wave-ripple-complex”) that is observable in the hippocampus. Because these LFP events can be detected quickly, they can be disrupted as soon as they start through a short electrical stimulation of the hippocampus.¹¹⁰