Here is a quiz. Let’s say someone says, “Sit,” to a dog, intending the word as a cue.
- What part of speech is the word, “Sit”?
- Then, what part of speech is the same word if we say, “Good sit!” afterwards?
That was a trick.
If we were talking to a human who speaks the same language we do, the first “Sit” could be an imperative or command verb. The second “Sit” would be a noun.
But neither of those, while grammatically correct, applies to training a dog. Dogs are not humans. “Sit” is something else entirely to them.
In dog training based on positive reinforcement, “sit” is a discriminative stimulus. To the dog, it is not a word. It is not English. It is not eligible for grammatical analysis. “Sit” is an antecedent, in this case a specific sound that comes to indicate that reinforcement is likely available for the act of sitting. (I include the word “likely” because sometimes we don’t reinforce every single sit.)
Examples of other discriminative stimuli for dogs are hand signals we give them, auditory cues such as whistles, and all sorts of things in life that act as cues that certain behaviors will be reinforced. These life events are not necessarily deliberate actions by us, and may not even be known to us. I wrote about some in my post called, “16 Behavioral Cues That I Didn’t Train (But Are Still for Real).”
So if “Sit” is a discriminative stimulus, what is “Good sit”? I’ll get there. First, I need to talk about this problem with words and meanings.
I Can’t Get It Out of My Head
We humans have an enormous problem to overcome when we use words as cues. When we hear the sounds that comprise the word “Sit,” in whatever language we speak, we can’t divorce the meaning of the word “Sit.” We generally pick verbal cues that are descriptions of the behaviors they apply to. Convenient for us, but unfortunate for the dogs. We can’t help but think they understand the cues as language. Then there’s the opposite problem of having a mythology about the word “no,” and arguments about whether it can be a cue or not.
Sometimes we pick more colorful words for cues for our amusement or because the standard word is inconvenient. My friend Marge’s cue for her dog Zip to sit is, “Senta,” the Portuguese word for sit. “Sit” was too close to his name, plus she didn’t want to spend his life sputtering out, “Zip, sit!” And although he’s a Portuguese Water Dog, she didn’t pick “Senta” because he innately understood it. He doesn’t. She picked it because it’s fun, clear, and didn’t resemble any of her other cues.
I have a couple of fun cues. I use “Yoga” to cue Zani into the bow position (downward dog, get it?). I use “Rewind” to cue Summer to do a funny little backwards crawl/scoot. But hey, I’m a human, so I still hear these as words, with meanings. Not just a group of sounds. (And of course, the “funny” part has to do with their meanings…just can’t get away from that, can we?)
The Curse of Knowledge
This inability to get the meanings of words out of our heads on behalf of our dogs is an intra-species example of the “curse of knowledge.” This refers to a situation where someone who knows something (in this case the human) can’t imagine not knowing it. Here is a link to a good synopsis of a famous study, “tappers and listeners,” about the curse of knowledge.
In the tappers and listeners study, one person in a team of two would tap out the rhythm of a well-known song. The other person had to guess the song. The listeners could guess correctly only about 2.5% of the time. But get this: the tappers predicted the listeners would know the answer 50% of the time.
The tappers heard the song in their heads as they tapped, and couldn’t put themselves accurately in the place of the listeners, who were only hearing tapping. Even the most empathetic of us can’t turn off the songs in our own heads.
In dog training, we are the tappers and the dogs are the listeners. It’s worse though, because not only do they not know all these meanings and subtexts that are there for us, they are not capable of knowing most of them. Yet they read situations so well and are typically so attuned to us that they give the impression of knowing these things in the same way we know them. They have their own geniuses, but it is not likely that any dog understands language and grammar as we do.
By the way, I am not the first to tell about the “tappers and listeners” study with regard to some characteristics of dog training. Kathy Sdao describes it and even demonstrates it in her DVD “What Not to Err.” My friend Marge incorporates it into her orientation for beginning clicker trainers.
“Good Sit!”
OK, I finally made my way back around to this phrase. You can easily find dozens of websites that instruct you to say, “Good sit!” after your dog sits. Probably some of you have been instructed to do that. I have. I was told the following by an obedience instructor: “You should say, “Good sit” after your dog sits so they will know what it is they did right.”
This assumes that the dog can follow the leap from “Sit” as a noise meaning that sit will be reinforced, to “Sit” as a noun, modified by “Good.” This makes no sense. It only makes sense in our twisted world where verbal cues unfortunately have meanings that correspond to the actions we attach to them.
Here is an example that I hope demonstrates the faulty logic of “Good sit.”
The starter’s pistol going off is a discriminative stimulus for people who run track. It indicates that pushing off the starting block and starting to run will likely be reinforced. So please envision this. A runner is practicing her starts. Today the coach is using a real starter’s pistol so she’ll get used to it. The coach fires the pistol, and the runner makes an excellent start. She runs a few yards, stops, and turns back. The coach says, “Good…” and BANG! fires the pistol in the air again. The runner startles and says, “Why did you fire again? I’m not ready! I’m not even in the block.” The coach says, “I was telling you that you made a good…” BANG! and fires the pistol one more time.
With this example, we can clearly see that that the cue is not the same as the action. The coach means to tell the runner that she made a good start. **BANG** is not a description of the action of start. It’s just the cue that indicates a certain action will be reinforced. Likewise, “Sit” is the noise that indicates to a dog that sitting will be reinforced. It does not somehow “mean” that action to the dog.
Frankly, I can keep this in my head only for short periods. It slides away so easily.
It’s Not Harmless
Some might say, OK, it doesn’t mean what we think it does, but it hurts nothing to say it, anyway. Well yes, there are worse things. But using “Sit” as part of a praise phrase is not a desirable practice.
First, you are repeating the cue when the dog is already doing the behavior. This dilutes the one-to-one pairing of the cue and the action, diminishing the power of the cue. It also adds more chatter to the training session, creating more verbiage for the dog to sift through to catch words that might be cues. Or to learn to ignore them. Finally, I believe we need to do everything possible to understand the dog’s point of view. Choose cues thoughtfully. Make sure they all sound different. Use them consistently, and only for that purpose. What if, instead of words in English (or your own native language), you had to use a randomly assigned color flash card or a complete nonsense phrase for every cue? Wouldn’t they be kind of hard to remember? That’s the position our dogs are in. They have to use brute memory on cues.
That last reason is the big one. Saying, “Good sit!” every once in a while or even regularly does little harm to the dog in the grand scheme of things. I’m sure I do half a dozen things to my dogs that are more confusing than that. The harm is to us as trainers. It keeps us entrenched in the belief that dogs understand language the same way we do.
If you are going to praise, far better to say, “Good!” or “Good girl!” or “Good dog!” And to say the same thing consistently. If you say it regularly before you give the treat, you are also building up a nice little conditioned reinforcer. But that’s a post for another day!
Related Posts
- “No” Is Not a Behavior…But That’s Not the Problem with Saying It
- Dog Training Basics: Getting the Behavior
- What’s an Antecedent Arrangement?
- Solving a Training Problem with Antecedent Arrangements
- 16 Behavioral Cues that I Didn’t Train (But Are Still for Real)
© Eileen Anderson 2015
Thank you!! This has always been a pet peeve of mine. I always imagine the dog is thinking: “I’m already sitting and you said ‘sit’ again. I must be doing it wrong.” (And then maybe they get up to try something else, but the human doesn’t notice because they are all done with that behavior!)
Exactly, julie! And great point about how we don’t notice. Already moving along in our heads…
Haha you always write such great posts, this has been a pet peeve of mine as well 🙂 Love the way you break it down and the “bang” metaphor.
Thanks Rose! I was hoping the “bang” thing worked.
Sorry, but you covered this so very nicely that I just have to say: Eileen, Good Post.
It’s been a long time since I thought about this, but a related set of questions was brought up years ago, on canine rehabilitation. That the dogs are often very new to learning about people and while we need to teach them things, we have no idea what commands and delivery will be used by the adopter. Thus the criteria for standard commands was set to commonly used English and common utterances, with multiple cues taught for some actions. Most of the dogs seemed to pick up on all this very well, even distinguishing between purposeful commands and those embedded in “human jabber” than can be ignored.
So, adding this to your post then prompts the question on a typical dog’s discriminative ability with phrases. As a sample of one, I called my dog over and said “Good Sit”. He said, “okay, that’s fine”, and did nothing.
Now, while I fully agree with your advice, I have to wonder, especially during initial learning, what percentage of dogs who are learning “Fido, sit” or just “sit”, would then be confused by a subsequent “good sit”. If I had to guess, I’d expect it to be greatly influenced by how the phase is delivered, but past that I don’t know.
Thanks Gerry!
I agree with your comments about the nuance. And you are giving me an excuse to say something else that I just didn’t have room for in the post. There are two situations in which I will deliberately say a cue that the dog is already doing. One is when practicing cue recognition. Saying “sit” while the dog is sitting (in which continuing to sit or grinding their butt obviously into the floor are both correct responses) for working on cue recognition. But also–for two of my dogs, I never trained a Stay cue. So if they already happen to be sitting, uncued, and I want them to sit/stay, I have to say “Sit” again. I’m consistent about it and I think they get it. This is actually a drawback of not using “Stay.” It’s pretty convoluted. To be absolutely sure, one could get the dog up and put them deliberately in a sit, down, or whatever.
You can see why I didn’t put this in the post….
Yup, I do see why you waited! And when dealing with only your own dog, I feel any sequence that supports effective communication would be good.
But now you went off topic a bit into the debate on cue association during learning, especially with capturing where it’s not used until the desired behavior begins. However, I’ll wait for your future post on that one.
As an aside, I started using “thank you” and “please”, and have gotten some interesting looks from people. The latter was from the question that, when a dog fails to perform a well known command, can a general “please” cause him to reconsider and perform the prior command with no additional prompting. The results thus far indicate that anything that prompts an attending response followed by a frozen wait on your part will do as well. But, “please” just sounds nicer, and helps to prevent people from then saying “sit, sit, sit, sit…”, and “thank you” prevents “good sit”.
Please and thank you are fascinating ideas!
English can be quite confusing! I was watching my Lab roll around in the dirt and duff the other day. He stood up and and started ambling towards the back door. Yikes! I said ‘Shake! Charlie, Shake! He stopped, looked at me kinda confused like and then sat and offered me his paw. Boy, did I ever feel dumb.
Good boy, Charlie!! I hope you are proud of him despite feeling sheepish. That’s great that he responded correctly to the verbal when you weren’t giving the “right” context cues!
Thank you for sharing this knowledge, takes away another of my mistakes
I’m glad it’s helpful! Thanks for the comment.
“First, you are repeating the cue when the dog is already doing the behavior. ” Are you really repeating the cue? The cue is SIT, not GOOD SIT. My guess is that a dog hears these two differently, as you indicated in your post (saying “good sit” doesn’t elicit a sit). So why isn’t “good sit”, like “good girl’ just a secondary reinforcer? If a dog distinguishes “sit” and “good sit” differently, then it might be verbiage, but wouldn’t necessarily be confusing to the dog.
As well, in watching dogs respond to 2 or more syllable cues, it seems to me that the response is to the first syllable. Such as in OK for a release. How many dogs wait until the “K” to release? Not many that i’ve observed. They anticipate from the “O”. For those that use this release word, try saying O-L or some other combination of O and a second letter and see if your dog releases. Assuming that the first syllable is the most important, then “good sit” would not be (much) different than “good girl”. And because good girl is usually a secondary reinforcer, then good sit would be the same.
I agree with your part about inflection. That’s kind of where Gerry was going, I think. They probably don’t hear it as the same thing.
That’s a good argument about the secondary reinforcer. Though in real life I have not seen any overlap in the people who say “good sit” and the people who are carefully and deliberately conditioning secondary reinforcers. Also if one is saying “good sit” and “good down” and “good target” –why not just, “good” instead of counting on the dogs only heeding the first syllable? I do have at least one cue with multiple syllables where my dogs are listening for that last word, “Mat.” Maybe it’s because that’s the way I inflect it, “getonyour MAT.”
The main point for me remains my final one. It’s not so immediately bad for the dog. It’s bad for handlers who think that the dog is understanding it as language instead of sounds. And there can be fallout for dogs from that. Just like there is dominance in nature, but most of us flinch the second we hear the word because of the BS it gets attached to in some dog training camps, and because the BS seems so “sticky” to the general public.
Thanks for the thoughtful and well argued comments.
jarah’s mom, I’m going to retract my argument about how the people this post might be addressing would not be conditioning reinforcers. It’s not relevant to whether my argument was good or not. Red herring. I rewrote this post yesterday to focus more on the global effects of the assumption, rather than possible harm from saying, “Good sit.” You have made a good point about how the statement could in some cases by pretty benign or positive. You’re right. It could.
When testing for word/context recognition by burying cue words in otherwise neutral sentences, very often the cue word did not prompt a response unless it was either the first word, or emphasised by tone or volume. If this was preceded by the dog’s name before a sentence, the next word following retained its effect. All of which supports your observation, and also Eileen’s example.
With additional practice on embedded cue words where the emphasis is gradually decreased, many dogs will learn to pick up on the embedded cue word.
DANG IT! Now I wish I’d used “rewind” for Jo’s backwards commando crawl rather than “creep”.
Now to figure out if I want to mess with a working cue just because *I* like yours better…..
Heh heh! I stole that from Marge Rogers! It’s just too good, isn’t it?
I have long found this practice of saying “good sit” to be one that makes no sense to me — I always perceived it as pointless because it presumes that dogs understand the concept of adjectives. Your interpretation that, in addition, dogs would need to understand the difference between “sit” as an imperative and “sit” as a noun essentially speaks to the same objection and is one that I hadn’t thought of, but that I truly appreciated! Thank you for articulating so well why this practice is one that should be avoided if we are to communicate effectively and clearly with our dogs.
“Good sit” drives me absolutely crazy as it dilutes the conditioned response of the command “sit”
When we say “sit” we want the dog to sit. When he hears “good sit” he is already sitting so he cannot do the action of sitting, thus diluting the word sit as a command.
The scenario changes when on says “good stay”. I believe the dog ignores the good part and hears a reinforcer of the stay command. Personally I still wouldn’t use it.
(Another point is the sit good or is the dog good for sitting… yup, Good dog!)
Love that last point, Martha! Thanks for commenting.
Hi Eileen, I’m new to dog training and am enjoying your blog. I have a dog and a parakeet. The parakeet is a male and enjoys talking, and I have clicker trained him as well (he actually came into our lives before the dog and started on clicker training before the dog did!). The parakeet has a few commands, such as, “up,” and “perch.” The funny thing is that when he is enjoying free time outside the cage, he often performs the actions of “up” or “perch,” and he says the commands as he is performing the actions. It seems to me that he has internalized the commands and knows that the action and verbal cue are essentially equivalent, if that makes sense. Considering that parakeets do not have formal structured language like humans, and are fortunately blessed with a talent for vocal mimicry which dogs lack, what do you think the consequences of such an example are? Do you think that saying “good up” or “good perch” is understood by the bird to mean that he did something well, since he already knows that “good” is a reinforcer? (The parakeet often says the words “good bird.” I believe that he knows that this phrase is a reinforcer.) I am leaning toward the opinion that you are still correct and the bird cannot understand a phrase such as “good perch” as anything other than, “reinforcer followed by next cue.” I think it is an interesting example, either way, and I am looking forward to your response! 🙂
I can only say, “Wow.” This is beyond even speculation for me. But it’s fascinating! I’m so glad you shared the story but I can’t even guess what’s going on there. Hopefully some other bird folks will chime in. I love this!
Excellent post Eileen, thank you