Fascinating question.

I think if you want precise timing in a web browser then that will be tricky. And in general you really don't want to get into the business of trying to support multiple web browsers, all of which operate slightly differently (and I suspect you don't really want to become a full-time web developer).

For my part, I'm a big fan of matlab/pscyhtoolbox, but I'm increasingly using Unity, as it has native support for smartphones, tablets, and VR.

If there is some way of combining Unity with Google's new game streaming service (Stadia; announced this week), then I think that might be a very strong option for you. I believe Microsoft are also launching a similar streaming service soon. That could be very effective for behavioural studies, though there will still be a non-trivial latency that would be a big red flag for any imaging.

Unity can also export to a browser-based app (search WebGL, or see https://www.ucl.ac.uk/~smgxprj/vr/ for an example). This is promising as it lets you run OpenGL graphics in a web browser (which is what you need). Again though, the frame-rate and functionality varies markedly depending on browser, and trying to maintain it long-term might not be completely trivial. Potentially worth exploring though.

I'd also note that OOP, when done right, is a wonderful thing. Maybe worth giving it another crack.

Good luck with it and do feedback with whatever direction you end up going, pete

Hi Pete,

thank you so much for your detailed answer. Unity does sound interesting indeed, I’ll take a look. I tried opening your link with iOS Safari but was not able to do so, does it require the subject, in this case me, to download a plugin?

Re streaming service: but wouldn’t it be better to run it on the local university servers to avoid GDPR hassle? Would be interesting to hear your view or maybe I misunderstood you.

Thanks, Annika

The webgl link I gave might just take a long time to load (i think by default it downloads the webgl code at first run). Or maybe it's user error on my part. If you google WebGL I'm sure you'll find much better examples than mine.

I don't think GDPR compliance should necessarily be a decisive factor. Whatever platform you use, there is nothing to stop you transmitting the data to a secure server/database for storage (e.g., my host institution -- UCL -- provides staff with a mySQL server from which I can push/pull data). A 'gold standard' application wouldn't typically store any data in the user's device, or in the file space used to host the app, wherever it is hosted and however it is coded, so some sort of database is a must (if one wants to maximally GDPR compliant).

I think for now the most used and most-tested approaches are JavaScript based with Canvas for stimuli drawing. In terms of performance, it's quite good, human noise is magnitudes larger. So if you don't need precision up to 1 ms, and can tolerate an occasional skipped frame, you'll be fine. When I did my studies online, I used to write the code myself (I mean, without experiment-specific stuff), but there are now handy libraries, like, for example, JSPsych.org or lab.js. PsychoPy also allows to port to Javascript (check out pavlovia.org) but AFAIK it's pretty limited for now.

Thanks everyone for your replies, really helpful!!

From my point of view, a good option is to use JS and framework Meteor.JS - it is a tool that allows you to implement all the functionality from the site as a whole and the database to individual experiments created in simple JS by different experimenters. In this case, typical blocks (such as checking the speed of a computer and browser) are must be done once and then used in new experiments. And it is good for people who do not have a technical education (but are able to master the programming logic). I’m a psychologist, but I mastered these tools and with my colleagues we did a series of experiments (for example, https://testpsy-aws.herokuapp.com/1/603 https://testpsy-aws.herokuapp.com/1/701 https: / /testpsy-aws.herokuapp.com/1/702 https://testpsy-aws.herokuapp.com/1/703 - they are in Russian, but you can see and understand the meaning of the tasks). Now I continue to work in this direction on a common system of experiments, and I like these tools, I plan to use them for the next research.

Also in javascript is jspsych (www.jspsych.org). For my stuff, I use javascript with p5.js for graphics, sound and input.

That you all, this is very useful input! Will check these out.

A primer on running human behavioural experiments online, Tijl Grootswagers https://psyarxiv.com/wvm3x

Thanks, Valeriya Azorina! I have settled on PsychoPy in the end, but this could still be an interesting read.