March 1, 2005Is the Brain an Analog or a Digital Device?
The once-great debate over whether digital is better than analog is over. As everyone knows, digital won. Sort of. Perhaps it would be more accurate to say that digital has won as a public-relations tool. The word digital has become synonymous with clear, undistorted, and high quality. Some of you may suppress a private chuckle when you see "crystal-clear digital sound" on the package of some cheap gizmo. Certain members of the professional recording community quietly resist digital manifest destiny by continuing to record on reel-to-reel tape, an analog medium. Nevertheless, no one can seriously disagree that digital media dominate in the home.
In the end, recording media are used for the purpose of delivering music to the brain. They are part of a kind of musician/brain interface. If the recording-medium end of that interface is digital, then what about the brain?
Lets review the pros and cons of digital and analog.
Digital media encode information in discrete steps that can be exactly represented by numbers. A sound-pressure wave that looks like a smooth curve on a pressure vs. time graph, for example, can be encoded as a series of numbers representing pressure and time. If you sample the real wave six times, the digital data will look like a staircase with six steps. Sample it a hundred times and the data look much smoother, because they comprise a staircase of 100 steps. Theoretically, you can sample the real wave as often as you like, and make your digital representation as close to the real thing as the capacity of your storage medium allows.
Analog comes from the word analogous, meaning having similar properties or behaving in a similar way. An analog representation of our soundwave might be an electrical voltage that varies with time in the same way that air pressure varies with time in the real wave. Hence, a graph of the electrical analog wave would have the same shape as a graph of the real wave; i.e., no staircase. Analog media are physical representations whose resolution is limited only by the accuracy with which the microphone and other equipment captured the sound.
The downsides of analog communication are noise and distortion during transmission. If your electrical wave gets distorted or contaminated, you can use filtering to control noise to some extent, but filtering tends to add its own distortions, and any distortions that get by the filter are nonrecoverable. For this reason, copies of an analog recording are never as good as the original.
A digital representation, on the other hand, is more robust in transmission. Signals that represent numbers can be off by a little bit and still be correctly interpreted. Further, multiple levels of error correction, such as checksum bits, can be incorporated into the medium, allowing near 100% accuracy of transmission. Media that incorporate such error-correction algorithms can therefore be copied with no loss of quality in the copies. This is true of the CD-ROM format, for example, where bit accuracy is required for software to run correctly. It is not true, unfortunately, of the CD music format, which was not designed for bit accuracy. Hence, copies of a CD are never quite as good as the original. Newer formats such as SACD and DVD-Audio, however, can be copied bit-accurately, a fact that makes content providers highly uncomfortable.
The brain, of course, must handle representations of information as well. When you get past all the mystical mumbo jumbo, the brain is a computational device that must make use of the only two information-handling formats we are aware of: analog and digital. So which is it? Actually, the brain is both.
Looking at the hearing system as a representative example of the way the brain works, the cochlea (which looks like a seashell, and whose name means seashell in ancient Greek) of the inner ear contains a ribbon-like membrane of decreasing width called the basilar membrane. The tension and width of this membrane are such that its resonant frequency gets higher the farther along the membrane you go. Sounds transmitted to the cochlea from the eardrum, or tympanic membrane, by the little bones (ossicles) in the ear make the basilar membrane vibrate in a position that corresponds to the frequency of the sound.
A series of sensors called hair cells are arranged along the basilar membrane. Like other cells in the body, the hair cells have a voltage difference of around -70mV between the inside and the outside of their cell membranes. Like other sensory cells, the hair cells are specialized neurons capable of producing a spike-like change in their cell membrane voltage, called an action potential. When the basilar membrane vibrates, the hair cells nearest the vibration respond by generating an action potential. The action potential is an all-or-nothing, self-regenerating spike in membrane voltage that travels along the axon of the hair cell within the auditory nerve to the brain. Because it is self-regenerating, the size of the voltage spike does not diminish with transmission distance.
This has all the hallmarks of binary digital transmission. If theres an action potential, thats a "1"; if theres no action potential, thats a "0." Slight distortions in the size or shape of the action potential caused by, say, radio interference, are still interpreted as a "1," so the information transfer is robust. The presence of an action potential signals the presence of the frequency monitored by that particular hair cell, while the number of action potentials per second signals how intense (loud) that frequency is.
Once the information reaches the brain, the axon from the hair cell divides into multiple branches that connect, or synapse, with neurons in the cochlear nucleus, a multifunctional group of neurons in the brainstem. Some of these neurons are concerned with frequency information, others are concerned with sound intensity, and still others compare the sounds received in one ear to those received in the other, looking for small differences in arrival times that give clues to the direction of origin of the sound.
Action potentials arriving at a synapse cause little packets of chemicals called neurotransmitters to be released into the gap, or synaptic cleft, between the incoming axon terminal and the target neuron. The neurotransmitters diffuse across the gap and act on chemical receptors on the surface of the neuron. The neurotransmitters released and the receptors present on the other side are specific to each synapse. Depending on the synapse, the action potential will lead to either an increase in the membrane voltage of the target neuron (excitatory transmission) or a decrease in the membrane voltage (inhibitory transmission). The change in membrane potential is local, and peters out with distance from the synapse.
Note that, at this point, the information is no longer digital. The disturbance in the target-neuron membrane voltage is not an all-or-nothing phenomenon, but a smoothly varying (continuous) wave that depends on the rate of arrival of action potentials at the synapse. The wave of voltage change on the neuron follows the intensity of the frequency detected by the hair cell and is analogous to it. In essence, a digital-to-analog conversion has occurred at the synapse.
It is rare for a single synapse to be strong enough to cause a neuron to fire (i.e., produce another action potential) all by itself. At the same time as our little synapse is acting, hundreds of other synapses both inhibitory and excitatory, from the ear and other regions of the brain, are also releasing neurotransmitters onto the same neuron in different locations on its surface. For the neuron to fire, the sum of the effects of all of these synapses must result in a net increase in membrane voltage above what is called the threshold potential. (The membrane sits at around -70mV "at rest"; the threshold for firing is around -40mV.)
If the neuron "decides" to fire, another action potential will travel down its axon toward its targets in a digital manner. Within the neuron, however, the decision-making process has all the hallmarks of analog processing. Each synapse produces a local disturbance of the membrane voltage that physically (not numerically) sums with other disturbances happening at other places and times on the neuron. These disturbances, or signals, are continuous rather than 1s and 0s, and any distortion of them will not be correctable. Fortunately, they are traveling very short distances on the surface of a single neuron, so distortion in transmission is minimized.
To summarize, nerve fibers in the brain transmit data digitally, but the neurons themselves do their computations in the analog domain. The conversion from digital to analog happens at the synapse.
Assuming that the wisdom of Mother Nature exceeds that of Intel Corporation, what is the advantage of analog computation? As the early engineers who built them knew, analog computers can easily perform operations that are hideously complex when done numerically, such as integration. The brain must constantly work with time-variant phenomena, for which integration is fundamental, so this is a major advantage. Analog computers also lend themselves well to parallel processing, another feature for which the brain is famous. Integration calculations can be difficult or even impossible in the digital domain, and parallel digital designs are unwieldy to execute.
What does all this mean for your stereo? Well, lets see. In home audio reproduction we have digital storage and transmission of data, followed by a digital-to-analog conversion in the CD or DVD player. From there an analog signal is amplified by the amplifier and transduced through the speakers, both of which are analog devices. As in the brain, information-transmission duties are handled digitally, then the signal is converted to analog and processed in analog. Mother Nature would approve.
Ultra Audio is part of the SoundStage! Network.