I was asked 6 years ago how I keep up with new trends and technologies in web development. My answer was simple: I'm constantly doing new projects, and the research I do to solve new problems is what keeps me updated on the latest tools for the industry.

Since then, I've included a few other items in my daily routine. It's worth it to peak in on what's happening at Hacker News, and I also follow a few web development and java development twitter accounts. 

When I want to learn something new and I don't have a direct project in mind, I'll read a book on it and follow along with the exercises. Other times, I'm just skimming for information so that I can broaden my existing solution set.

By far, though, the best way to learn something new is to do something new. Don't make it a goal to "learn to code" -- make it a goal to make a video game, and then learning to code will just become part of it. I'm not trying to learn 3D modeling; it's just one part of the next steps I have to make my next game.

Once you're done with a project, start a new, bigger project. Go from 2D games to 3D games, as I'm doing. Go from learning piano to writing a symphony. Or start something different altogether. Either way, the learning process is best bundled up with the overall project itself.

There are times when general research is necessary and enjoyable. Understanding how something works for its own sake is a fun leisurely activity. Sometimes, all I do is "research". Right now, I'm learning Latin. That's considered "research" to me, as it isn't directly related to anything I'm working on, even though I do plan on using it in my next game to an extent. I will also do "research" on existing technologies that come up in things I work on at work, just to get a bit of background into the subject.