Emotient's emotion-recognition technology derives from the Machine Perception Lab at the University of California at San Diego and has focused primarily on helping advertisers understand viewer reactions to their ads.
The acquisition was first reported Thursday in The Wall Street Journal, but terms of the deal were not disclosed.
Just a few months ago, Apple acquired another AI startup, called VocalIQ, whose speech-recognition technology is widely expected to smooth interactions with Apple's Siri virtual assistant.
Now, Emotient's technology could bring emotion recognition to Apple devices, though the company is notorious for keeping its plans to itself.
"It's almost impossible to figure out Apple's complete motives -- even after an acquisition is made public, it tends to play its cards close to the vest," said Tuong Huy Nguyen, a principal research analyst with Gartner.
That said, the Emotient acquisition also seems related in some ways to Apple's acquisition of German augmented-reality startup Metaio last May, Nguyen suggested.
In essence, both use algorithms to deliver new kinds of intelligence, he explained, with potentially significant benefits for users.
If the next iPhone could recognize frustration in users' expressions as they navigate a particular sequence of steps, for example, the aggregated data could give Apple early insight into the need for design changes, Nguyen pointed out.
There are clearly privacy issues associated with facial-recognition technologies like Emotient's, but if managed carefully, the benefits could outweigh them, Nguyen said.
"Privacy is something that needs to be handled very carefully," he said, "but using algorithms to process data about the world around us could make our devices smarter and easier to use."
Emotient's website has been stripped of much of the description it once offered. Neither Apple nor Emotient could immediately be reached for confirmation or further detail.