Premeditation makes the difference between first- and second-degree murder, and being convicted of the former tends to bring much harsher sentences. The article describes the paper as finding that people who view (e.g. store CCTV) recordings of a murder only in slow motion and without a timecode, thus lacking any clear reference to the real speed of events, are much more likely to impute premeditation than those who don't. That's what "distortion" means here.
Think of the ultimate slow motion, a photograph. Emotions and muscle movements sweep across a face, but depending on the photographer's or editor's choice of moment, the same person can look scary or beatific.
If A intended to kill B, that's murder.
If A did not intend to kill B, it's manslaughter, a lesser crime with lesser punishment.
You are sitting on the jury, with the prosecution flashing between two frames of a video of the punch and arguing that A altered the trajectory of the punch to cause B to fall onto the hard surface.
Two frames is, say, 33 milliseconds, or somewhere between 2cm and 4m (I admit, I was surprised by that speed) of neural pathway, taking the lowest and highest nerve conduction velocities I can find. How many intentions and motor changes can you perform in 33ms?