Liking and Attention

This past year Robert Heath has presented, on a number of platforms, research which he claims shows "THE MORE EMOTIVE AN AD …
THE MORE YOU LIKE IT…THE MORE YOU TRUST IT…THE LESS ATTENTION YOU PAY TO IT"
This continues the theme he has plied for nearly ten years now; but the research is new.
He starts with the understanding that eye movements are ideal for measuring attention levels. Tiny totally autonomic rapid reflex movements, known as fixation speed. This is an approach which has been used successfully with static stimulus, such as print, in the past. But he uses it for TV ads. And it is from this work that he concludes that the more you like an ad, the less attention you pay to it, and "High levels of emotional content in advertising were significantly correlated at 99.9% with LOWER levels of attention"
This fits perfectly with his theory, yet flies in the face of all the other research into this topic.
Why should this be?
I don't know. But - I suspect the answer lies in one sentence of his papers (e.g.‘KRUGMAN WAS RIGHT – TV ADVERTISING GETS LITTLE ATTENTION BUT BUILDS BIG BRANDS’) . "A few subjects started by watching the screen carefully and followed the action, but most watched in a ‘lazy’ way, exactly matching Krugman’s description of ‘motionless, passive eye characteristics’."
In other words, the thing he was measuring, eye movements, stopped being measurable a few moments into the ad, because the eyes became motionless. So the only way he could distinguish between the levels of attention paid to different ads was through the responses picked up at the start of each ad.
So, if Heath's sentence is an accurate summary, we have the inference that his assessment of the level of attention paid to the ads was actually a measure of the attention paid to the first few moments of the ad.
Can this be right? Surely if this were the case, he would have highlighted the fact in the papers? But none of them refer to it.
There is one other factor here worth reporting on. In his most recent paper, he refers to the r squared for the relationship. This is a measure of how close the relationship is; an r squared of 0 shows there is no relationship between the measures, an r squared of 1 shows there relationship is perfect. In this work, the r squared is 0.104. It is significant - but tiny. And, arguably, what you might expect if your measure of attention was based on the first few moments of the ad.