During the manipulation check, I found that dogs were more likely than chance to approach the tunnel containing the transparent screen first. This provides evidence that the dogs remembered the different properties of the tunnels from the experience phase and that they needed to alter their position to see the object of interest, in this case, the experimenter chopping the sausage.  The manipulation check was therefore successful in testing the dogs’ memory of the properties of the tunnels. 

I checked to what extent dogs would engage in gaze following when the experimenter looked towards a target object. Dogs’ looking time and first look at the bucket behind the tunnel containing the transparent screen was significantly longer, and more frequent when the experimenter looked at the tunnel containing the transparent screen compared to the opaque screen. Conversely, I saw a similar trend though it was not significant of dogs’ looking times and first look at the bucket behind the tunnel with the opaque screen when the experimenter looked at the tunnel containing the opaque screen compared to the transparent screen. This result shows us evidence of gaze following in dogs since the dogs followed the experimenter’s gaze.

I also compared the gaze congruent looking times to see if the dogs were more likely to follow the gaze, in the transparent condition compared to the opaque one. Dogs’ gaze congruent looking times were not significantly modulated by the looking condition. The result indicated the dogs did not interpret the experimenter’s gaze cues as informative about the location of hidden food, because they looked the same amount to the two tunnels. 

I argue that this experiment fulfills the requirements that skeptics proposed to validly test mentalizing skills in nonhumans. Because the dogs did not experience any associative learning during the experience phase. During the trials, the dogs could only discriminate between the tunnels visual properties based on their previous self-experience and use this knowledge appropriately to infer what the human experimenter could see. Hence this experiment is a first step in investigating the attribution of mental states in dogs in a novel context and disentangle mentalizing from mere behaviour-reading skills.


Dogs look longer and look more often first towards the bucket behind the transparent tunnel when the transparent screen is cued compared to when the opaque one is. This provides evidence of gaze following. 

The results do not provide clear support if pet dogs can use their own experience with novel barriers to infer if someone else can or cannot see through the same barriers. 

This can be explained that dogs just do not project self-experience onto others or due to some limitations of the methods.