When is the reinforcer not so reinforcing?

Kay Laurence
 Jun 2008

As trainers who live in the world of reinforcement it becomes our responsibility to explore the depths of reinforcement and discover the underlying effects of continual use and misuse.

Initially we struggled to learn about the four quadrants: positive reinforcement (R+), positive punishment (P+), negative reinforcement (R-) and negative punishment (P-), and then discover  Professor Jesús Rosales-Ruiz was set to confuse us with a 5th “quadrant”: extinction. Not many people are aware of which quadrant they are controlling from second to second, and often jumping very fast between each area, even change over between the click and the reinforcer. The food comes out of the treat bag, is lying in anticipation (R+), and then not delivered due to a perceived failure. Or we click for response, and then hesitate in the delivery of the food with several seconds of silence. This can punish the behaviour and the next repetition will slightly slower, the dog now has doubt.

Keep it as simple as possible and continually remind yourself that if the behaviour is maintained or getting stronger then it is being reinforced (R+). Even if that is not your intention, intention doesn’t change behaviour. If the behaviour is fading or stops then it is being punished (R-) . Even if that is not your intention either. Make a habit of looking at each incident and assess what is happening to the behaviour. The evidence can only be gathered from the dog performing that behaviour at that specific time. There are several underlying complexities that have an affect on the behaviour and the first incidence or change in response may not be the true picture, you may need to set up several repetitions and look for a trend in the changing behaviour.

A good reinforcement history in association with the behaviour affects the chances of that behaviour maintaining its quality on the next repetition. But if the reinforcement history is poor, or so varied it has caused uncertainty, the behaviour will fade or extinguish a lot quicker when reinforcement is not forthcoming.

What is reinforcing?

Although each recipient can vary, and vary between different behaviours or environments, there are obvious reinforcers that work most of the time: such as food or attention, and very subtle but important reinforcers that occur more often or are more continuous. These can be used to enhance the obvious reinforcers, or negate them if used carelessly.

In one situation a dog’s behaviour can attract you away from another focus, it will likely reward the behaviour. The dog can start to chew furniture or start barking when you are on the phone, or watching TV. The behaviour interrupts you and your attention is refocussed onto the dog. If the attention is positive: you fuss the dog, talk to it, or fetch the guilt-treat, you are now stuck in a cycle of reinforcing an unwanted behaviour that is trigged by you ignoring the dog. Useful huh? Even if you shout at the dog, or move it away, for some dogs the gain of your attention is more reinforcing than any punitive association of you being cross.

In a class situation I see dogs regularly learn to trigger owner attention, without them noticing the strategy. A behaviour that you may perceive as embarrassing will catch your attention. The dog wanders off, sniffing the floor, and you move towards them, catch their attention, cue their name. They often get double rewards: attention and a treat. Easy. It made you focus on them. It is going to happen again. The dogs learn the pattern, they see your chin beginning to lift and begin the chain of lost attention.

What we perceive as a reinforcer may be a punisher in different situations. A dog may love physical contact when relaxed, around the house or just being in your company. But that same gesture may predict the end of a walk, capture, or the loss of a good game when associated with the lead being put on, collar grabbed, or pushed into cage. Now contact is a predict of the loss of something reinforcing. 

Delivery of food can also have a twist. If the dog is in a high state of arousal or anxiety, eating the food may not be that pleasant. The dog may take the food because of the reinforcement history associated with the action, but it may be hard to swallow on a dry mouth, or sit uncomfortably in the mouth or throat.

Reinforcers that are derived from fixed actions patterns, or using instinctive behaviours, are very reliable reinforcers when used in appropriate situations. But asking a dog to grab onto the tug, and give 100% attention may not be reinforcing when the dog is the centre of attention of a predatory stare. That behaviour may be the one step too far that arouses the predatory dog. Even though the dog may respond to the cue to “tug”, their instinct is telling them to play cool, relax, and this conflicting cue could be punishing – or even end in an attack. That will certainly stop tug games being reinforcing around other dogs for the future.

All the fixed action pattern range can be used as reinforcers: hunting, tracking, scenting, stalking, pointing, chasing, biting, gripping, ripping, tugging, de-gloving, carrying, swimming. In this category we must also include social approval. Nearly all our dogs successfully integrate into our life style because they have inherited a need for social approval, and comfort in belonging to a social group. This is triggered in youngsters by attention from senior dogs, disapproval or lack of attention. The patterns become reinforced and stronger.

In my time I have met less than a handful of dogs that had no desire for social approval, or human attention. How much was inherited and how much was extinguished would be hard to tell, but they all made teaching an uphill challenge.

Social approval is also labelled “praise”, but more effective when carried out as canine social approval rather than the stereotypical head slapping and “good boy”. I don’t enjoy the dogs delivering a tongue mouth wash, but do allow copious amounts of skin licking, leaning on me, holding my clothing and running around yapping.

Subtle reinforcers can be equally powerful as the obvious reinforcers since they are more frequent, and several can happen at the same time.

You can be carefully watching the dog whilst they puzzle out a new behaviour, suddenly you end your stillness with activity – and this was triggered by something the dog achieved. You cue an established behaviour in a tricky situation, and on success run joyfully to the fridge and deliver several treats. The treats are great, but the dog will be hugely reinforced by their power to activate you from stillness into a frenzy of actions. Wow, reinforcing huh?

The dog goes to respond to the cue and sees your hand move towards the treat source, (hand creeps to pocket or pouch), this causes hesitation, which is reinforced by the anticipation of the reinforcer. Anticipation is reinforcing – try the same technique before you give the cue and notice the change in the behaviour. I’ve even seem a dog go to respond to a cue, look at the treat bag, cueing the owner to put their hand onto the bag before they respond to the behaviour. And dogs are equally skilled at cueing trainers to give them a click. They complete the behaviour, raise their eyebrows: “huh? So then …. Where’s the click?” Sure enough the click and food follows – good reinforcement for the behaviour of cueing the trainer! Oops.

If a behaviour is always associated with a particular reinforcer, when it is carried out that reinforcer is anticipated. Delivery of a lower value reinforcer could punish the behaviour. Whoops, the dog may let you make the mistake once, but just may not come back on recall the next time to get a head-slapping Good Boy next time around. We use anticipation to build a reinforcement history. Some dog will even begin to salivate where they hear the behaviour cue, or get aroused because a high action game is the usual reward. Beware disappointment. Plan the reinforcement pattern carefully.

Choice is reinforcing. The whole process of teaching yourself, making safe choices with the security of some guidance, is fantastically reinforcing. Add onto that your exclusive attention (until Kay wanders over to join in), your arousal when they make an achievement AND food. It’s called free-shaping. BUT, a big whopping BUT, too much choice or choice with no guidance can be punishing. Look at food menus or supermarkets – too much choice can bring us to a stand still, but limit the number and we can still feel the reinforcement. Just think how reinforcing choosing the reinforcer could be?

What is punishing?

Dogs find our attention reinforcing. They learn as puppies to enjoy the focus, the arousal, and get that wonderful warm feeling from being the centre of attention. Unfortunately as they progress through the teenage months they are less appealing and will often discover – through single event learning – that obnoxious behaviour is needed to get you to notice them: jumping up, peeing everywhere and anywhere, pulling on lead, crotch sniffing, ignoring their name etc.

In class people regularly punish wonderful behaviours and well mannered teenagers by ignoring their good behaviour and turning away to talk to other people. Yeow! As I approach the owner is naturally drawn to talk to me (either to tell me the excuse of the week, or seeking personal reinforcement) and as people we are taught to make eye contact when talking. Don’t look at me!

How many repetitions must the dog experience to learn that an approaching teacher will commence the withdrawal of attention? Of course the behaviour will deteriorate as I wander over. Don’t look at me!

I see many competitive trainers set their dog up at the commencement of a test, then when they are ready and the dog is demonstrating the perfect behaviour, they turn away from the dog and talk to the steward or judge. Yeow!

The loss of a reinforcer is punishing. This is obvious when the reinforcing is continual. Shaping sessions are highly enjoyable for Mabel, and I used to end the sessions with a pat, close the food pot, “that’ll do” language. For a dog that is very tactile in every day life, that end of session pat predicted a punishment and began to look like I was delivering an electric shock: “don’t you touch me!”

Turning away from a dog often represents withdrawal of attention. Many about-turns in obedience heelwork are subtly punishing. No wonder they become traumatic: the dog hears the cue “let’s do an about turn” – “close” and the owner promptly turns their back. Eeek, I would wonder how I would cope with that? If you are in conversation with another person what are their subtle signs that punish your behaviour? Looking at their watch, texting, focussing beyond you, partly turning away, leaning away. The dogs have been studying your behaviour all their lives, they have Honours Degrees in You, so of course they can see they subtle cues that you are no longer focussing on them.

The space in front of you is a place with a massive reinforcement history. This is Manager’s Slot No. 1. Whatever is in that space can manage your behaviour, can block your way, force you to manoeuvre and trigger your attention. Being in that space becomes highly reinforcing. You may need to remove access to that space when an undesired behaviour starts. The dog is playing tug and “accidentally” bites your fingers. Even if you abandon the toy, the dog remaining in the Manager’s Slot No. 1 is reinforced. Not only must attention be withdrawn, but access to that space denied. Move to somewhere that prevents access – face the wall, a chair or other person.

Getting the balance

My gift to you would be the magic wand that slows the world down to a pace where you can anticipate what is coming, think about whether to reinforce or punish, work out which is in place, change your behaviour and reassess the result. Yeah. Panasonic invented videos just for us. Don’t be afraid to video your session and observe all the R+ and R- that are happening.

A balance must be achieved between avoiding reinforcing the undesirable behaviours, and this may be as subtle as the dog’s loss of attention, floor sniffing or room gazing, and avoiding punishing the desirable behaviours: great attention on you, ignoring other dogs, staying under control. This imbalance alone will seriously impede your progress and can fool you into wrongly interpreting the outcomes.

Set “self awareness” days. Find one element that can be either a reinforcer or punisher: gentle attention or eye contact from you to the dog. Make sure this does not resemble a predatory stare. Focus through the day on using this in all interactions until you become aware of turning it on or off. Over breakfast, if the dog is present and settled, make eye contact, if the dog is pestering you for end of toast focus on the tea pot.

Make yourself a list and practice your skills: become aware and conscious of the subtle changes that over the long term have a significant effect on your dog. If you are entering competitions or attending classes find a technique to explain to people around you that you are doing the exercise: Dog Comes First. You will acknowledge them, but maintain focus on your dog whilst you require their focus on you. People will still talk to you even if you don’t make eye contact (unfortunately!).

I have a reputation of being an unsociable sod when I’m with my dogs. My relationship with them supersedes all casual relationships with other people, and especially in the competition environment when I expect them to stay connected to me, I will stay connected to them.