The debate is still alive, but more researchers have, like James, now take the position that both sides can be true.
Making Faces Can Increase Body Muscle Strength
Several researchers have proved that grimacing actually increases hand strength. Making a face produces a direct effect on a seemingly un-related part of the body. Just as grimacing intensifies grip strength, making strong facial expressions can intensify experience of other emotions.
Recent computerized assessment of multiple site Facial EMG activity has demonstrated its superiority over observer visual assessment of subject's emotions.
During the past 10 years many studies have repeatedly shown high correlations between facial muscle activity and emotional state. Fair and Schwartz reported that normals show stronger zygomatic response during positive affective imagery. Depressed patients exhibit stronger corrugator responses and weaker zygomatic responses. This seems analogous to the pattern physical therapists encounter when using biofeedback to rehabilitate weakened or atrophied muscles. One muscle (like the zygomaticus) is underactive. The antagonist muscle (the corrugator) is overactive and must be voluntarily inhibited and controlled.
We have been using Zygomaticus activation and Corrugator muscle inhibition EMG feedback for positive affect facilitation, intensification and "smile rehabilitation. Prospective data is being collected. When subjects are instructed to maximally activate the zygomaticus, readings range from 12 microvolts (100-200 hz bandpass) to 150 microvolts. Practice appears to dramatically increase contraction strength above initial levels. When subjects are induced to laugh or smile naturally, their EMG activity tends to be higher than during volitional efforts at maximal zygomatic contraction, or even maximal efforts to smile.
This suggests an inhibitory process at work, perhaps similar to what occurs during the early stages of thermal biofeedback training, when efforts to produce vasodilation usually result in cooling of the fingers. Further zygomaticus increase training, coupled with biofeedback monitored smiling and laughing to facilitate subject "connectedness" with the awareness of psychophysiological dimensions of positive emotion seems to lead to the ability to equal and then exceed automatic or "involuntary" positive emotional response EMG activation.
Caccioppo mentions one Japanese study in which a group of human cadavers were dissected and two percent of them were found to be lacking their zygomaticus-- the primary smile muscle. Were they atrophied through lack of use or missing from birth.