What is the "Optimal" Virtual Reality Experience/Future

"Markystal" wrote:
Fascinating insight Astrocyte. The nature of qualia, the expansion of sympathy, the possibilities are endless and I certainly hope that a metaverse or cyberspace can be used to help people truly interface with each other with genuine communication of feelings rather than our current methods that can be misconstrued.

I certainly hope so to. But, it will largely depend on how we implement and (where appropriate) monitor such a metaverse/cyberspace construct. Career trolls, whilst a minority, can cause a disproportionate amount of trouble for everyone through their relentless bullying and abuse (these are often people with a serious personality disorder). Just look at the problems this forum has had with "trolling" over the Easter weekend. Now imagine this happening in VR. The potential for psychological harm with VR is very real, and we all need to consider how people can be best protected in the future.

EEG whilst useful in some contexts, has several issues that optimization and computational improvements can only take us so far in solving. Your right that mind reading using EEG is impossible. As a measure of post-synaptic neural activity, the strongest advantage to using EEG is that it has excellent temporal resolution (i.e. millisecond-range resolution which is not possible with CT or MRI). For this reason derivatives of the EEG technique such as Event-Related Potentials (ERPs), which involve recording neural activity time-locked to the presentation of a stimulus (i.e. visual, somatosensory, or auditory), are useful for answering certain types of questions in cognitive neuroscience research. However, such responses need to be averaged over many hundreds of trials, and this takes time and significant computational resources to accomplish, so yes, there is latency in this regard. The main disadvantage of EEG is its poor spatial resolution. Even though in most cases, high density arrays are essential if one is going to have a sensible estimate of where particular signals are coming from in the brain, there is always the problem that voltages spread out across the scalp and thus, become blurred. Using computational algorithms can correct for this, but only so much. In addition, the best systems also need a small amplifier in each electrode in order to help with the problem of ambient electrical noise. Your correct that slight movements of the head can create artifacts in the data. I'd also add to that jaw clenching, eye-blinks and movement from the body generally. All these cause artifacts that usually need to be removed during the pre-processing stage before one can get useful data to play with. I don't know much about using EEG, EMGs, EKGs, and EOGs to track the occurrence of particular body actions, but it would be interesting to try. But, I would expect the signal to noise ratio to be a variable thing depending on what the actions were, what environment one was in, the degree of adipose tissue (fat) and, what type of electrode setup one was using. The ambient temperature may also be an issue as sweating often causes changes in skin conductivity. Processing wise, I really think it depends on how well today's basic quantum computers scale up in the longer-term. This could potentially take us beyond Moore's law, but currently, hard evidence is scarce in terms of the relative advantages of such systems over conventional computers. I'd like to think that in the future, we could harness the computational power of biological neural networks (perhaps even our brains) for rendering virtual environments, but that is some pipe dream at present! ;) . I agree with Palmer that input devices will need to allow two way real-time communication. This is a significant challenge when one considers brain machine interfaces, or indeed muscle/nerve rift interfaces.

TMS is an interesting approach to brain stimulation, but, its effectiveness varies across people. One factor which affects the fidelity of TMS is differences accross people with respect to cortical folding patterns. Also, currently most TMS systems cannot hit deep into the brain, so their application is restricted to outer cortical layers. I am aware that people are experimenting with different shaped electromagnetic coils, but not sure how these fare in terms of stimulation depth. As I don't see technology surpassing biological neural tissue for some time, I agree that at this point, I'd rather stick with non-invasive technologies! Thinking of how people will use this stuff rather than stuff itself is of value. For real immersion I suspect that one will have to find ways of dimming awareness of the body and its immediate environment. But, as you point out, most people don't want to be burned to death in a house-fire whilst in VR! So, in very immerse environments good monitoring via A.I. is going to be central to safely, although that is a big ask since multiple fail safe protocols would need to be present before most people would even consider it. And, what if you cheese that A.I off? I agree that VR on the go is unwise, but AR through the rift may be a different matter if it is sufficiently un-entrusive. For that to happen though the rift will have to shrink to little more than eye glasses, and, I don't see that happening anytime soon.

"cerebral" wrote:
The problem with nerve interfaces is obviously the surgical expense.

Its not just a matter of expense in so much as what is also technically possible, particularly with non-invasive technologies. For most A.I., surgical or not, human supervision is likely to stay a significant feature for some time to come I think. Even then, I suspect that many people will prefer to be treated by a human, unless of course it becomes impossible for most people to know for sure that the "person" they are talking to is not human....

Fascinating discussion from both of you. Thankyou! :D