Animal Welfare Professionals

Updated 10/2 at 2:03pm - Recording now available to watch on-demand!

We hope to see you on Monday, 10/2 at 11am PT for our next Community Conversations call. We'll be joined by Dr. Federico Rossano, Founder and Director at Comparative Cognition Lab at UC San Diego as he presents the research on the use of button pressing in dogs and cats which help animals communicate their needs.

Once the idea of a talking pet was restricted to children's books. Today, pet owners are endeavoring to train their dogs, cats and other animals to communicate through words. In this talk I will introduce you to the largest community science project ever undertaken in the field of animal intelligence, spanning thousands of participants from 47 countries and including several social media stars such as Bunny the Talking Dog. I will showcase how soundboards with world-labeled buttons can be used by animals to express their needs, wants and feelings, and how we are studying this phenomenon scientifically. We believe the findings of this study have the potential to ultimately change how we interact with animals, improve their rights and mainly help protect their welfare.

#Behavior,TrainingandEnrichment
#DataandTechnology

------------------------------
Maddie's Pet Forum Admin
Maddie's Fund
------------------------------

Unanswered question from the chat, 

We have connected with several researchers that are part of the interspecies communication community quite early on (among others, Dr. Irene Pepperberg and Dr. Diana Reiss, who have been trailblazers in this field). Indeed we have presented some of the plans for our study at a large interspecies communication conference in summer 2021 (https://www.youtube.com/watch?v=SluQJZ0gtOk&t=242s I start talking around minute 4) and we have a paper that is going to appear in a special issue on interspecies communication and technology edited by Dr. Irene Pepperberg. 

This said, I confess that I am not familiar with "intuitive interspecies communication" but I'll be happy to look more into it. Thank you!

Unanswered question from the chat, 

"Could this be useful for support canines?" - Christine Best

Yes, that is our belief. Or better, we strongly believe that any tool that improves communication and provides more choice and control over the animal environment is bound to have beneficial effect on the animal welfare. Note that we are not suggesting this type of tools should replace more traditional dog behavioral training or any research dedicated to understand canine spontaneous communication. That is all critical. We think of these tools as the equivalent of learning a second language. You might not need it most of the times, but sometimes, while interacting with people from a different culture, it might come in handy. Moreover we believe these tools could provide researchers with a novel window into canine's minds. Much more research is needed, of course. But once you look at data coming from thousand of participants, some interesting things become apparent and we have not received any complaint that training with buttons has ruined the bond between humans and their dogs. Rather, we have only heard of a perceived strengthening of that bond.

Unanswered question from the chat,

"How will the research identify or control for anthropomorphic inference on communication and delineate that from the intent of the dogs' communication?" - @Lawrence Minnis

This is a very important issue that would deserve much more than a short answer in a forum. Anthropomorphism is indeed a recurrent issue in Animal Language studies and animal communication research in general. This is why I mentioned the research we do with several other animal species, especially non-human primates. First and foremost we are trying to connect whichever phenomenon we are looking at with what we currently know about such ability in other animals through observational and especially experimental work. We know for example that dogs can learn several words and appear to understand what they mean. Indeed there have been several scientific papers published on the matter. So we start by carefully reviewing the dog cognition and communication literature, the literature focusing on training animals with AIC devices (see for example our review paper here: https://wires.onlinelibrary.wiley.com/doi/abs/10.1002/wcs.1647) and we make sure to adopt several sanity checks so that we are not fooled by random reporting. That is why for example we look at the reporting that what most dogs are doing with their buttons is requesting for things. That is indeed what we expect to be the case and if the dominant answer was "sharing thoughts" or "talking for others", I would have been concerned. But when you have dozens of people reporting on such other things then you are forced to take a look and figure out ways to investigate such behaviors further. So as mentioned we do have 24/7 chronic recording on some participants to capture potential cueing, we run behavioral experiments with control conditions (and research assistants run experiments in people's homes to further reduce bias and cueing)  and we make sure to pre-register all our studies, so that other scientists  can check on what we intend to do. We also plan on share all the data we have collected as soon as the publications begin to come out, to allow the general public to see the evidence we have for what we are claiming. We also rely on ethograms, developed for dogs and cats, so that we can capture non-verbal signals that most likely are critical to our understanding of what is going on.

Overall, what we try to do is be first and foremost aware of the risks of anthropomorphizing these animals and we want to keep as close to what is biologically plausible as possible. But what I also want, as a cognitive scientist, is avoid the psychological bias we have adopted for centuries, where we think that We alone, as humans, have the ability to think and communicate about complex thoughts or feelings. If nature does not make jumps, then a continuum is more likely, and the idea that animals that have been selected to live with us for thousands of years might share with us some impressive cognitive abilities might not be too far fetched. Until science tells us otherwise, of course! I am a hardcore empiricist and so all I can say is: let's look at the data as thoroughly as we can and let's see what we can make of it!

Unanswered question from the chat, 

"What type of buttons are being used? We tried some and the sound was not good enough for the dog to understand what was being said." - @Penny Leisch

I am sorry to hear about the problem with the sound. We usually try to avoid specifically advertising brands because we want to maintain independence and we welcome participants with any type of buttons, no matter what brand they are. This said, it is publicly known that our project started as a collaborative project with FluentPet, a company that produces buttons and they have recently put out  a model (Connect) that allows for automatic logging through an App and with a centralized speaker that makes the audio more intelligible. But then again, I am sure there are many other buttons that can work. 
What we have heard is that size of the buttons and how hard one has to press can be a problem for some breeds because dogs come in so many different sizes and the buttons that work for a Terrier might not necessarily be ideal for a great Dane. You can find more info and contact us directly here: https://cclab.ucsd.edu/dog-cognition-project/ 

Unanswered questions from the chat, 

"What doesn't make sense for me is the ones where the human has the camera all ready to go, and then the dog walks over to the board. Like they knew in advance what the dog was going to do?...I mean the ones where they are holding a cell phone up! Not the security camera types" - Maria S

    "is the camera motion activated when the dog goes to the buttons?" - Sharon Fletcher

Excellent questions and concerns. As I mentioned in the talk, the clips  I showed at the beginning of the talk are taken from social media accounts and are NOT part of the data we analyze for our study. I have shown them as examples of phenomena we are interested in, mainly because they come with nice subtitles already ;) The last clip I have shown, where we run the impossible task experiment where the dog asks for help, is actually from one of our studies and not from social media.

I feel the same way about many clips on social media. They often seem cherry picked or possibly staged or simply represent the one time out of 100 where things seem intelligible enough to make sense. This is why we collect our own data and look at automatic logging of all button presses, recordings from cameras that are running 24/7 in some people's homes and experiments that we conduct ourselves. This said, yes, several people in the study have their own camera setup (these days there are several cheap security camera options) that are motion activated and focused on the soundboards. So they are not dependent on a human turning on the phone and starting an interaction but rather record anything that is happening. 
I am with you on this: the recordings from security cameras seem more "real" and less staged. But I guarantee that our claims and what you will read in our scientific papers are NOT based on social media clips. It would not make sense otherwise. 

Unanswered question from the chat, 

"can deaf animals use buttons?" - Lisa Wetherby

This is a great question! We do have several blind pets in the study but I do not believe we have deaf ones (but I'll check with our team and correct my answer if I am wrong). In principle they should be able to learn as well the contingency between pushing a button and something happening in the world. But we do know that the sound does help, because when the button is broken and there is no sound the dogs and cats are surprised, keep pressing the button and they generally do not bark while pressing the buttons, supposedly because they understand that humans need to hear the sound for it to work. 
We do believe that they memorize the buttons through spatial mapping and the symbols on them (and the hextiles the button might be on). But we have not specifically focused on deaf pets. If you have a deaf pet and would like to be part of the study, please reach out to us. You can find more info about our study here: https://cclab.ucsd.edu/dog-cognition-project/ 

Unanswered question from the chat, 

Yes. Several participants report carrying the tiles and buttons with them while they travel and we have had some social events where participants in the study meet to socialize and bring their buttons and their pets. Then the soundboards are set on the floor in somebody's garden or yard and the animals interact with them and their humans while also interacting with each other. This tells us that the mapping of meaning to button is not just dependent on the location of the soundboard in the house but rather is retain also while the soundboard and the buttons are relocated in a  new environment.

This is why we believe that shelter dogs could be train on a few buttons and then that soundboard and those buttons could be moved to the household of the human adopting them and this would still work.

Unanswered question from the chat, 

"Have you trained 2 animals together where they use the buttons to communicate to each other???  That would be interesting." - @June Thomas

Yes. Or better, the research team is not training any animal directly. This is a community science project. Several participants have several pets in their homes and they are training them with the same soundboard. Some have reported having the pets using the buttons to communicate to each other at times (especially between pets belonging to different species like dogs and cats) but some have reported it also for two dogs for example. We need to look more closely at this because at times it seems to be more about communicating with the human/making the human aware of what is happening. But it is potentially very intriguing. 

Unanswered question from the chat, 

"Are they selecting by location when there are 100+ buttons or do they find the right button when the board is moved or rearranged, which would imply recognizing the letters on the buttons too?" - @Penny Leisch

We have not done a systematic study concerning how they recognize the buttons but we believe it is a combination of factors, starting from their spatial location (e.g. on the edge of the soundboard) within a frame (it could be the tiles) to the symbols on the buttons. We do know that the animals get surprised if the sound coming out of the button is different from what it used to be (e.g. if you switch buttons or record a different word/phrase on a button previously in use). So it is likely that the symbols on the button help them. From watching videos of dogs with more than 100 buttons I can confirm that they seem to know pretty clearly which buttons to push and there is not a lot of hesitation. 

Unanswered question from the chat, 

"Is there consideration for Pavlov's law with a button and then reaction/reward vs communication with no reaction/reward associated" - @Laura Zmud

If I understand the question correctly, it concerns the degree to which what is going on is basically operant conditioning (i.e. the animal gets to learn a button because of being rewarded when they use it). We do believe that operant conditioning is most certainly a crucial part of the learning process (and incidentally this is true for humans as well, not just pets).  We are looking into the different learning strategies, from classical conditioning where we compare for example participants that might learn by being exposed dozens of times to the contingent occurrence of a button press and some outcome (e.g. pressing outside and then going outside) vs. learners who push the new buttons themselves and then look for what the human does next (operant conditioning). We clearly believe that conditioning matters. What we are trying to assess is to what degree, once a concept has been learned, the animal can flexibly use it and combine it with other concepts in novel ways. Something that should be impossible through classical or operant conditioning and through pure association.

Unanswered question from the chat, 

"I would love to hear more about when you would consider a more abstract word/feeling to be trained? Like love, concern, etc." - @Heather Kalman (She/her)

This is a wonderful question that is particularly tricky to answer. The short answer is that one can look at this the same way we can try to assess when a child seems to understand an abstract concept like numbers, love, etc: if they use it in a contextually appropriate way and their response is contextually appropriate, then for all practical purposes they understand the concept. In general, all traditional training with words and commands has gone towards teaching them what to do (actions) and for some teaching them names of objects, and we know that names of objects are harder than names of actions (contrary to human children). So several people have serious doubts that a dog could ever understand the concept of "concerned" or "love". And I agree. But I also have a 17 month old that tells me "I love you" and a 3.5 year old who tells me "I love you" and a wife who tells me "I love you" and I am confident that their representation of what "love" is differs enormously among them... but ultimately would be satisfied with a hug and a kiss back. Does it mean that my children do not know what "love" is? One could say so, but once again, they seem to be using this expression at the right time in the right context and as far as I am concerned, it works. 
We often believe that humans who share a language have the same representation of the same concepts/words, but we now have scientific evidence that this is not quite the case. An excellent paper was published this year in Open Mind (https://direct.mit.edu/opmi/article/doi/10.1162/opmi_a_00072/114924/Latent-Diversity-in-Human-Concepts) showing us exactly the level of divergence across people in terms of how they represent the meaning of different concepts. So the real question is: if the representation differs among people, when can we say that the concept has been properly learned? It is an exciting  question worth working on.

Unanswered question from the chat, 

Not yet. I would LOVE to collaborate with them.

Right now detection dogs are trained to detect things and we ask them to signal to the human when they have detected it, usually by stopping the search and sitting near the location where they have detected something. But they cannot tell us exactly what they have detected. Imagine if they could tell us what else they have detected or simply exactly what they have detected and where it might be located. I believe it could make dog searches significantly more precise and efficient, which would be critical while searching for explosives or missing people.

And one would not need to train for dozens of words. Just a few would be most likely sufficient. And military and police dogs are by definition highly trainable. 

Unanswered question from the chat,

"Yes! What are implications for promoting/justifying the value of dogs to those who reduce them to 'just dogs'" - @Karina Reyes

A previous question has reminded all of us that we should not be anthropomorphize the animals. Yet we often are dealing with people who go exactly the opposite direction: they treat dogs by definition as irrelevant agents. As not worthy of care, love, attention because of their perceived lack of humane features. If we can show that they have emotional and cognitive abilities comparable to a toddler/young child (this is still an empirical question of course but not a completely crazy hypotheses for anyone who has ever interacted with a dog), then the public would be forced to consider what their rights should be. After all, how would we treat a child in a similar situation? What would we think of a person who abandons a child? Or somebody who hurts a child or deprives them of care? While this is a scientific study aimed at unraveling the cognitive and communicative abilities of our pets, there is a long history of scientific findings that have ultimately affected policy, especially with respect to how to treat animals (e.g. because they can experience pain). So we do hope that any insight we can gather will hopefully have positive implications for animal welfare and animal rights.

Unanswered question from the chat, 

"Implications for sending our dogs to school to learn how to "speak"/communicate?" - @Karina Reyes

Well we do send our dogs to get trained for several things, so learning a few words to more easily communicate their needs and wants to their humans might not be such a crazy idea , especially if the ultimate outcome is a stronger bond and reduced frustration on both ends. Let's see what the ultimate findings from the study are, but as an idea it might not be very far fetched!

Very interesting talk yesterday! I taught my dog to ring a bell to go potty--I'm sure lots of people have done this. But, duh (!), it never occurred to me to use more buttons and teach her other things.  I see that the effectiveness of button board communication relies on operant conditioning principles, whereby a reward is given to the dog in order to get the dog to repeat the behavior in the future.  In the absence of bacon bits in the dog videos we saw, I assume that the "reward" for dogs is getting what the dog is asking for--water, to go out. In addition, I have to also assume that there is reward in the mere communication--getting your attention, you talking to them and interacting with them. Where I was impressed/confused/a bit skeptical is with two scenarios--1) where a dog asks for water for their pal and 2) where a dog "expresses" concern. I am very curious about how these "skills" were taught.  I get 'press the button' for water or to go potty--that seems straightforward.  But how was the emotion of concern taught to or "trained" into the dog? Yes, we can give a dog a lot of credit and say the dog is genuinely concerned. But is it possible that we are anthropomorphizing the dog's behavior and that it did not reflect actual concern?  And with the dog that wants water for their canine pal, I wonder if the dog advocating for the canine roommate also drinks from that bowl and was simply asking to have that bowl refilled so they can have water rather than have it refilled out of concern for their buddy.   Was there a starting point for what concern means and how it is identified as concern? In other words, what is the operational definition for concern? How do we get to call something "concern"?

These are all excellent "concerns" :) and indeed something we are trying to be careful about.

I realize that I might not have been super clear in my presentation about the nature of the clips compared to the nature of the research. Those clips are NOT the data we are using to make any scientific claims. They are just easy to use examples in presentations to convey the type of phenomena that participants in the study have described, shown, raised and that we are trying to monitor through different means. In other words, the clips are absolutely anecdotal and several alternative interpretations are possible. I would be a very bad scientist if I claimed that the clip in which Flambo asks for water for Espeon is by itself clear evidence of Flambo caring for the other dog living in the house (note though that Espeon goes to drink water when water is provided and Flambo does not touch the water bowl, hence the explanation that Flambo was asking for water for himself does not quite work here). What I am rather saying is that out of the thousands of participants in the study, a few dozens have suggested that one of their pets, the most proficient one with the soundboard, appears to be using it to ask for help for other pets in the household. I have seen myself at least another dozen clips from different households where pets do things that on the surface seem to convey that they care and are trying to help. For example I have seen several clips where one dog is stuck outside or behind a door that suddenly closed and the other goes to the soundboard calling the human for help, often mentioning the name of the other pet. The one pushing the buttons does not move away from the buttons in those cases and indeed the human realizes that the help is not for the button presser, who appears completely fine, but rather for the other one, barking behind a door. Is this clear evidence that they care for others? probably not, and that is why we need to collect these instances systematically and possibly replicate these scenarios in behavioral experiments. But they resonate with research conducted for example with rats showing them opening a restrainer for another rat to let them out, research published in Science (https://www.science.org/doi/abs/10.1126/science.1210789) While we can debate whether a rat might experience empathy, we have scientific experimental evidence showing that dogs can produce empathy-like responses to distress calls from other dogs (https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0152920) so I am not sure it is too far fetched to believe that they can be "concerned" or experience empathy for their fellow pets in the house when they are in distress. 
This said, I agree 100% that at this stage of our knowledge we cannot trust that a dog pressing "sad" or "mad" or "concerned" is actually experiencing the emotion that humans would label as such. Hence we need more research. But we also know that when I say "I am sad" and you say "I am sad", we have most likely different representations of what "sad" means and we have no access to what the other is actually experiencing. It is the label we were taught as children when we were experiencing specific situations in the world. To put it differently, the current state of research on emotions in humans would not warrant any evidence that 2 humans have the same representation of "concerned", though to some degree those representations most likely overlap enough to allow us to communicate and make sense of each other. 

So how do we learn emotions as humans? Not by operant conditioning but by having caregivers label what we are going through and developing a strong association between the physiological experience and the label we have been provided. 
The way the participants in the study are training their pets with buttons is very similar to how we would teach language to a child. By using the buttons to describe what is going on and/or announce what is going to happen. We call it "modeling".  And over time the pets start using the buttons and the humans produce appropriate responses that count as the reward. Both classical and operant conditioning are at play, but what we care about the most is not how they learn buttons (though it is important) but most importantly how they use it. So if they tell the human "belly ouch" and then a few minutes later they puke... we have reason to believe that they were talking about actual pain. And if they say "ear ouch" or "paw ouch" and we see that indeed they are irritated or there is something stack in their paw, they might be talking about pain that they are actually experiencing. 

While we should not anthropomorphize the pets, I agree with you 100%, we should also not end up on the other side of the spectrum, somehow denying the very possibility that they might experience and be able to communicate about emotions and pain and distress and be responsive to it. 

I welcome all the help and advice that we can get and skepticism is the most important ingredient of science. Lots of what we currently think might be the case will likely turn out to be a fluke. That is why we are careful about what we can actually claim scientifically and the evidence we have for it. But I do believe that if we start with the assumption that "dogs" cannot experience empathy or be concerned for other dogs... we might be missing out on a chance to discover something quite fundamental about them. We should be careful and skeptical. But we should be open minded. or at least that is what I am trying to be ;) 

Hello,

I just had an opportunity to watch on demand, and I was impressed!  I kept thinking about how the animals must feel very important to be able to communicate with humans this way.  I think it is also a huge relief to them in so many ways, especially with the healthcare piece.  I can imagine how frustrating it could be at times trying to tell us what they need.  My dog not only uses a bell to tell me he needs to go out, he uses the bell to tell me he wants a treat.  This happens when he rings the bell and walks into the kitchen until I follow him to where the treats are, then sits and waits for me to give him a treat.  When he asks me continuously, I end it by saying, "All done."  He also uses the bell to tell me he wants me to go outside with him.  This happens when he rings the bell and stands by the door until I walk out with him.  Who is training who here?  Because he is so intelligent, I think he would greatly benefit from this kind of training.  I will gladly pass the information on to our shelter.

------------------------------
Laura Sherman
Volunteer
Mary Ann Morris Animal Society (M.A.M.A.S.)
Bamberg SC
------------------------------