Which approach is right for Machine Learning?

Basically we have two ways of approaching Machine Learning.

• One is by trying to build brain inspired algorithms. Say, Neural Networks Deep Learning and Cognitive models which aims at mimicing brain functions in understanding and solving data.

• Other is influenced by Statistics which has widely adopted language R and has methods like Bayesian inference.

• Even Michael Jordan has adopted Statistical approach over Cognitive. And also has said in in his recent interview that Neuron is basically a cartoon and we are not doing it right in getting close to understanding our brain.

• Whereas Jeff Hawkins of Numenta has come up with Cortical Learning Algorithm which is much inspired by brain functioning.

So what is the right approach which one should consider?