Really, this is hardly a surprising result. There are many possible mechanisms that suggest themselves, operating either on the embryo or on the newborn - parents who are more intelligent are likely to be able to pass on more of what they've learned and/or provide a "richer" environment for their offspring, even if we're only talking about mice. The mammalian brain is remarkably plastic.
The real problem for the Lamarckian paradigm is that once you've optimized the environment, socialization, and gene expression for the animals in question, it's hard to propose a mechanism for making more radical changes through "acquired characteristics" - and in fact no such changes have been observed. This study does not change that fact.
The original article sounds to me to be altogether too credulous and sensationalistic.