Can technology that prompts or inhibits neuron firing be able to add/alter brain pathways and behavior?

Question

Imagine I had a bunch of little molecules or electrodes or chips or something of the sort, which were outfitted with the ability to attach to brain cells, receive specific electric signal/field patterns from the brain activity, and in turn either prompt a neuron to fire when it normally wouldn't or inhibit firing when it normally would. These little devices react to both specific electric patterns, and to the activity of their neighbors, to coordinate and form larger patterns with the brain cell they're attached to as one "node" of that pattern. This function allows them to react to specific signal patterns that correspond to a certain sensory input, and form a new "reaction" within the brain to that input by generating that altered pattern of electric activity among brain cells. Basically:

1. Brain receives sensory input, sensory neurons signal that specific stimuli.
2. System of neurons with attached devices receive the signals from the sensory neurons. Depending on the signal an individual one picks up, it either has no active effect, prompts its neuron to fire when it normally wouldn't, or inhibits firing in a neuron that otherwise would.
3. This activity comes together into a new pattern of activity and brain pathways, creating a sort of "installed" electric-activity response to a stimulus, that normally is not present in the brain's behavior.

The devices themselves may work technologically, but if they are biological they may "determine" the response to a pattern by translating an electric pattern to chemical patterns internally, and then adding up the "sum" of chemical ratios and dynamically activating (to produce their electric/inhibiting effect) or remaining inactive in response to a particular sum.

Would this general system be able to work to alter brain behavior and "reactions" to particular input, or is there something I'm missing about neurons that would make this fundamentally impossible or too dangerous to do in this way?

Answer 1

People have put electrodes into brains and stimulated individual cells. They have made small sets of living cells do simple computations. At the lowest level, the answer is 'yes'.

The brain is plastic enough to adapt to new stimuli. You can add signals from an artificial retina, and the subject will learn to interpret the signals. Other experiments with monkeys given a third colour channel shows that adults can learn to incorporate a new signal with an existing one. If you are delivering extra information and letting the brain process it, then we can do that. For example, if you put current information from Google Maps into a series of electrodes, the subject might develop a sense of location and orientation.

If you think of trying to do some of the processing on the machine side, things are more speculative. Most neural network models reduce many signals to a smaller set, and the expand it up again. It is reasonable to suggest that the smallest layer in a trained neural net may represent the 'essential information' within the data. This would require the fewest electrodes, and might do some of the processing that the brain would otherwise have to do for itself. This is unproven, but possible.

We have no circuit diagram for the brain. We may get computer models that we can analyse and infer something about the nature of understanding. But scanning a brain and downloading the personality, or plugging a new part into the brain that works without having to train the rest of the brain to interpret the new signals is much, much further away.

Is the subject willing to have the new signals implanted? If this is done with their co-operation, they could train to recognise the new signals, and this could work. If you are trying to plug something in that alters their sense if reality without them knowing, that would be hard. If that is what you want, I suggest a hybrid approach: the user trains to a set of signals that interfaces to a web search (for example), then replace that signal with another one that relays false data.

Answer 2

What you are describing is, more or less, just adding more (slightly different) neurons to the brain. As you don't state how these "Technological Neurons" actually decide what to do and/or learn it's nearly impossible to state output.

As for your understanding (as of current neuroscience State of Art):

• there is no direct, one way, path from sensor stimuli to "recognition"; what seems to happen is brain generates patterns (Mirror Neurons, which are NOT single neurons, but specific groups) and those are "checked" against sensory input, if they are "compatible" firing is reinforced, otherwise it's damped.
• Mirror Neuron patterns are stable for a single individual, but are completely unrelated (at least no correlation is found to date) between individuals; guess is those patterns are built while training the brain; this poses difficulties deciding what neurons to stimulate in the first place (should be a training phase where real stimuli are presented, "normal reaction" is recorded and pattern is saved, on multiple nodes, for future usage).
• we have many billion neurons in our brain and deciding which to "override" beforehand is "not easy", unless you plan to have billions of your devices.
• if they are "technological": how do you plan to power them?
• if they are "biological": how do you plan to control them?
• wouldn't be simple to implant a "loudspeaker" in the inner ear so the poor sod "hears voices in his head"?