Finally, you can now hallucinate in virtual reality. A team from Sussex University has produced a VR program they’re calling the Hallucination Machine, allowing users to put on a headset and wander around the uni campus while everything wavers and shimmers and morphs into dog’s faces.

The program is the brainchild of a neuroscientist named Anil Seth. He’s the co-director of Sussex University’s Sackler Centre for Consciousness Science—a research team that hopes to unpack the mechanics of human consciousness. But as he told Wired earlier this year, getting consistent data on consciousnesses is tough as consciousness shifts around with stimulus. But by using VR to control this stimulus, Anil and his team hope to more reliably measure consciousness.

“It’s been a longstanding goal that came about when trying to understand, from a neuroscience perspective, the notion of presence, the question of how and why we typically experience things are being real, as in really existing in the world,” he told Wired. “When we start asking that question, VR becomes an important experimental tool.”

And so, the Hallucination Machine now reliably produces a psilocybin mushroom trip, allowing Anil and his team to examine how the brain differentiates between reality and hallucinations. And the benefit of using VR is that you don’t have to flood the brain with psilocybin, which obviously alters its neurophysiology and distorts the data.

But why does the Hallucination Machine feature so many dogs?

Because they borrowed their visualisation technology from Google. You might have heard of Google’s weird side project, Deep Dream, which uses neural networks—layers of computing systems that mimic the biological networks of animal brains—to create psychedelic images. Led by a Russian programmer named Alexander Mordvintsev, Deep Dream takes the kind of technology that allows Google Images to identify billions of images on the internet, but flips that technology so it also creates images. Mordvintsev tweaked the program so it became hypersensitive to things that weren’t even there, encouraging it to find patterns in all textures and gradients, and then spit out a visualisation of whatever it was seeing.

And what it sees, apparently, is dogs. Mordvintsev just happened to using data sets that had been trained to identify dogs, which is why Deep Dream ended up finding dogs in everything. And so when Anil Seth and his team wanted a program that could create some hallucinogenic imagery for VR, they borrowed the Deep Dream code, and got it to mess with panoramic video footage shot around the uni campus. And now anyone using their Hallucination Machine will see a vision of campus life if the very fabric of existence was somehow comprised of dogs.

The next obvious question: how will this be useful to science?

So far, Anil’s research team have used their psychedelic VR in two separate experiments. The first put 12 people in headsets and then quizzed them on what happened: whether they saw any colours and patterns, and whether they felt a loss of control. Interestingly, their answers corresponded very neatly with findings from a 2013 study on the experience of taking psilocybin.

The second experiment, however, revealed that the Hallucination Machine can’t replicate all the features of a psychedelic experience. 22 people were given headsets and asked to stroll around the virtual campus, before being asked whether their perception of time had been affected. Here, most answered no.

The results seem to suggest that while hallucinations can be replicated on a surface level, there’s a lot going on in the brain we still don’t understand. And basically it’s hoped that this trippy piece of VR will give researchers another tool to further understanding. As the report's authors conclude, "Overall, the Hallucination Machine provides a powerful new tool to complement the resurgence of research into altered states of consciousness."

(Original article)