It’s a strange object, really. Electrically-driven, composed of unimaginably small components, and quite connected to psychology. This description applies both to the brain and to Solid State Drives (SSDs).
The brain is unbelievably efficient (at least when viewed against hard drives). Comparing your run of the mill brain to, say, a 100GB SSD, it has about 1/8 as many neurons as the drive has transistors. This is with neurons being about 3 orders of magnitude larger. And yet the mind is capable of storing far, far more than a 100GB drive could ever hope to encompass. Additionally, power consumption is quite different. For instance, once something is stored on a drive, not a whole lot of energy has to be expended to keep that information stored. (Which is why a hard drive can remain completely unplugged for long stretches of time and not lose information.) This is much unlike a brain, in which an interruption of a couple minutes can have deadly consequences.
Yet the similarities are there, functionally speaking. This is why analogies between the brain and computers have long wormed their way into psychology, with information processing theory back in the 1950s drawing up an analogy between the brain and computers as its central tool for understanding the mind. Under that framework the mind operates like a processor, where it receives an input, processes it, and produces an output. This analogy (present in everyday life, frankly) is what motivated my original article’s title, “How Come Silicon Remembers?”
But this is where I admit that the first article’s title was a bit of a misnomer. The actual “memory” of Hard Disk Drives don’t contain much silicon, if any. Solid State Drives (SSDs), on the other hand –as the name suggests– are completely solid (no moving parts) and made in large part from silicon. But why bring them up at all? Well, if you’ve ever looked inside your phone –for instance– you probably didn’t see any spinning disks in there. This is because phones (and a vast array of other devices and technologies) use solid state drives, or flash memory. Over the last few decades SSDs have become hugely relevant to the world of digital storage.
But how do SSDs work differently than their HDDs predecessors?
To begin a distinction has to be made between volatile and non-volatile forms of storage. Volatile forms of storage (such as RAM) are kinds of storage that, when power is lost, immediately forget whatever was stored on them. Non-volatile forms of storage (such as a thumb drive, for instance) can “remember” things even without a constant supply of electricity.
How exactly does this work? To understand there needs to be a quick introduction to:
Transistors
Transistors are composed of three basic components:
The Collector: Where current starts, or “collects”.
The Base: What turns the transistor on or off, it’s a sort of “gate” if you will.
The Emitter: Where current flows to, or is “emitted”.
Transistors are, fundamentally, electrical switches. (Equally as important is the fact that they can amplify signals, but that isn’t wholly relevant to this article.) The issue is, though, that transistors also fall under the category of “volatile memory”. This is because unlike a physical switch that stays either on or off, a transistor’s default is off, so the moment the electricity is shut off all the transistors revert to their default state. Thus, transistors –on their own– can’t store information. To get around this SSDs also have what’s called a “floating gate”. Floating gates basically store a charge if the transistor was on, and are empty if the transistor was off; this allows data to be stored even if the electricity is shut off.
Beyond the architecture itself, transistors are made possible by the use of semiconductors. Unfortunately, this means that over time even drives with no moving parts (i.e. SSDs), have limited lifespans that can be measured in terms of rewrites, as the materials that make them up physically degrade. (Or how many times the floating gate has been charged and discharged). Modern SSDs have several adaptations to drag out their usability as much as possible, such as spreading out the rewrites across the entire drive, so it wears out at an even pace, but HDDs are still better for long-term storage.
This still doesn't necessarily make one better than the other overall, as they still serve different purposes. SSDs' primary purpose is speed rather than long-term storage. This is because, as said before, they degrade more quickly than their hard disk counterparts, but as a tradeoff they are also vastly faster. Relatedly, even with floating gates the data stored in the drive will start to degrade if left without power for long enough (around a year or two, depending on the specific drive). A sort of silicon-based asphyxia, if you want.
Now, how do transistors and neurons compare? There’s certainly some comparison, with neurons being in an “on” state (firing), or “off” state (not firing), and the reason that occurs is dependent on the inputs each neuron receives from the neurons around it (somewhat like an electrical circuit). However, there are also some distinct differences like the fact that in neurons being biological, they can also be impacted by things like dopamine or serotonin, which isn’t really something that is present in an SSD (analogously speaking, though also literally). Additionally, the brain is a lot more diverse in the way it stores memories. While hard drives distribute “memories” –regardless of their type– relatively evenly across themselves, brains are a lot more nitpicky with where things go.
For instance, not all memories are created equal. While a digital file’s type doesn’t really affect where it ends up on a drive, the same cannot be said of memories. Memories can be split into explicit memories (which can themselves be split into episodic and semantic memories), and implicit memories. Explicit memories (which involve the hippocampus, the neocortex, and the amygdala) are of specific experiences one has had, such as That Terrible Thing You Did –or That Incredible Thing You Did, if you prefer. Under explicit memories are episodic memories are recollections of some given experience, while semantic memories are a recollection of some given fact or piece of information. Implicit memories are basically just muscle memory (which unsurprisingly involves not just the muscles, but also the brain: specifically the basal ganglia and the cerebellum).
All in all, while it is definitely fascinating that a component of so much of our everyday technology acts —in some respects, like the very machinery in our own bodies— it is still only a limited comparison. However, technology can't seem to help approaching humanity. Look to neural networks, or machine learning algorithms. Those are much more comparable to the human mind, and could even one day give way to things like AGIs (Artificial General Intelligences), which do things just as well if not better than humans. So maybe the comparison stretches a little deeper than the simple comparison of two, independent things. Instead, the brain and SSDs, or AGIs, are intrinsically linked in the fact that we created it. One of the things being compared preceded and engineered the other. So instead, the comparison could perhaps be getting at how much of ourselves is reflected in the things that we make. In that case, it is no coincidence that psychology and technology are, at times, so very entwined.
Can we just stop for a moment and appreciate the fact that a part of our brain is called the “Temporal Lobe”? If we were talking about some sci-fi biological supercomputer the temporal lobe sounds like something that would meddle with Time itself. Unfortunately for us, it’s simply named that way because it’s next to our temples.
Comments