Sensitivity to Human Attentional States

By | July 23, 2013

One might expect that under circumstances in which deictic signals result in interference or exploitation by an observer that a dog might employ gaze amd directional cues to turn the others attention away from the location or object of interest. The results of a study by Call and colleagues () might be interpreted as evidence of tactical behavior organized to evade interference by adjusting risky ventures to changes in human orientation, proximity, and attention. In their study, dogs were exposed to training in which a piece of food was placed on the floor, whereupon the experimenter looked at the dog and said “Aus!” (Out!), followed by a second event in which the dog’s name was called and followed by “Aus!” again. The dogs were subsequently exposed to a series of test trials that continued until the dog either took the food or 3 minutes had elapsed. If the dog took the food without permission, it was not punished, nor was it rewarded if it refrained from taking the food. As such, the procedure appears to asymmetrically favor approach over avoidance, since dogs that took the forbidden food were merely ignored, just as those that obeyed the prohibition were ignored. With only one exception, all of the dogs took the food at least once, and some of them took food on several occasions while they were watched. However, watched dogs generally avoided the forbidden food but readily took the food if released to do so or when left alone with it. In a follow-up experiment, the dogs appeared to be aware of the attentional state of the experimenter based on the orientation of the experimenter’s direction of gaze. When the experimenter faced the dog with closed eyes, the dog tended to respond to the frontal orientation as it did when the experimenter’s back was turned to the dog. These findings suggest the possibility that dogs might know when they are being carefully observed. When closely watched, the dogs broke the prohibition against taking the food much less frequently, engaged in more indirect approaches to the forbidden item, and appeared less relaxed (i.e., sat more often and laid down less often) during the test trial. The indirect approach might have enabled the dogs to keep an eye on the experimenter while nearing the object or taking a more advantageous position from where to approach the food item.

A remarkable aspect of the experiment by Call and colleagues is the degree of appetitive suppression achieved by virtue of two vocal warnings, essentially events that emphatically served to draw the dogs attention to the food in association with a social startle/threat implication. One possible explanation for the extraordinary control exerted by vocal reprimands may stem from the unfamiliarity of the testing situation and the person delivering the threat. Another possible factor might be traced to prior inhibitory training. Some of the dogs may have received aversive inhibitory training that conditioned them to refuse food not given to them by the owner (e.g., poison proofing). Poison-proofed dogs usually show a strong avoidance and rejection of illicit food that they might find or have offered to them by strangers. In contrast, the dogs in the test group eventually took food when released or when left alone, strongly suggesting that they were not behaving under the influence of situational anxiety or aversive inhibition. Apparently, the experimenter remained neutral whether a dog ate the food or not. From a behavioral aspect, the effect of such a procedure would be far from neutral. In fact, such a strategy should have resulted in a significant differentiating effect with respect to dogs that ate and those that were inhibited. Dogs that ate the food in the presence of the neutral experimenter should have been doubly gratified by the disconfirmation of the avoidance contingency and by the successful control established over the food item without cost or interference. Learning theory predicts that such rewarding outcomes should reduce inhibition and conflict in dogs about taking food in comparison to dogs that continued to avoid the food.

Appetitive Suppression, Social Attraction, and the Attribution of Intention

Matters are further complicated by the fact that dogs tend to cope with social inhibitory conditioning in ways that are highly variable and strongly influenced by constitutional differences and prior rearing practices (). Freedman studied the ontogeny of these effects in several breeds. Puppies selected from these breeds were divided into two groups: one group had been reared with disciplinary handling and the other reared with social indulgence. At week 8, disciplined and indulged puppies received several days of inhibitory training during which they received swats on their rumps together with loud vocal reprimands whenever they ate meat from a bowl that was located in the middle of a room. The aversive contingency exerted a differentiating effect on social and appetitive approach behavior that was correlated strongly with the rearing breed of the puppy and only partially with rearing history.

After the experimenter left the room, basenjis, which showed a strong environmental and novelty orientation, tended to eat from the bowl straightaway, whether they had been punished or indulged, whereas shelties, which showed a strong social avoidance, consistently avoided the food during the full 8 days of testing, regardless of their rearing history. In contrast to the social aloofness of the basenjis and the social avoidance of the shelties, the beagles and wirehaired fox terriers showed a high level of attraction and interest in the experimenter. Depending on their previous rearing history, these breeds also differentiated into separate groups after punishment training. Indulged puppies showed a greater amount of avoidance toward the meat than did puppies that received disciplinary treatment prior to punishment. Of particular interest with respect to the long-term effects of such treatment, indulged beagle puppies receiving punishment training at week 8 showed evidence of delayed changes in social behavior consistent with a loss of trust and safety stemming from the earlier punitive experience. During follow-up tests from weeks 11 to 15, these indulged-punished beagle puppies became wary when approached by handlers and increasingly difficult to catch — behavior that sharply contrasted with their earlier sociable attitude and behavior. The so-called second fear period may actually be related to the retraction of residual social attraction or its loss as the result of interactive conflict, releasing conditioned social fear acquired earlier in life.

Freedman’s findings indicate that early rearing practices and individual differences can strongly influence how dogs respond to appetitive inhibitory training. The individual differences associated with appetitive inhibitions would seem to represent a significant confounding influence for cognitive studies such as the one performed by Call and colleagues. According to Freedman’s work, some dogs should simply avoid food in the presence of the experimenter, whereas others should simply ignore the warning and take the food at the first opportunity. Dogs most likely to differentiate behavior in response to reprimands would need to have a high level of social attraction.

Another problematic influence is prior safe or unsafe exposure to food. Under normal household conditions, dogs frequently find bits of food fallen to the floor, which they are usually permitted to eat without consequence. Most dogs are also accustomed to obtaining food as rewards from their owners or to obtaining it by searching countertops and trash bins — efforts that are intermittently successful and may persist despite the use of contingent and belated punitive efforts. Dogs regularly receive small portions of food as rewards consequent to the performance of obedience tasks. Food is also sometimes used to relax anxious dogs via its calming effects on aversive emotion and arousal. The idea here is that preexposure to food would likely build a significant level of prior conditioning and bias that should impede or facilitate the acquisition of inhibitory conditioning (see Stimulus Factors Affecting Conditioned-stimulus Acquisition and Maintenance). Further, there might be an intrinsic contrapreparedness associated with learning to avoid appetitive objects via aversive threats. Rats, for example, exposed to brief foot shock and other aversives (e.g., ammonia, mustard, or quinine) as they approached a highly preferred food item (an Oreo cookie), or as they ate the cookie, continued to show a persistent appetite for the object. In contrast, rats exposed to nausea-producing lithium chloride acquired a rapid and cross-contextual aversion toward the cookie, perhaps indicating a certain degree of independence between cutaneous and alimentary defense systems. When subsequently tested, the shocked rats showed only momentary hesitation before eating the cookie in the training cage where the shock took place but showed no hesitation before eating the cookie while in the home cage or when tested in a novel cage (see Prepared Connections: Taste Aversion).

Despite biological contrapreparedness, lasting food inhibitions can be established rapidly with severe physical or electrical punishment but not without risking neurotogenic elaborations. Lichtenstein () demonstrated that appetitive behavior could be fully suppressed by the delivery of one to three electrical shocks (2 seconds of 85-volt AC each), if the aversive events were timed to overlap the act of eating, whereas significantly more repetitions of shock stimulation (23 to 29 shocks) were required if it was delivered immediately before the presentation of food (see Lichten-stein’s Experiments), again raising additional questions about the size, durability, and ease with which Call and colleagues were able to establish a stable inhibition of appetitive behavior.

Chairs and Minds: On Knowing What a Dog Knows About Attentional States