|Home | What's New? | Cephalopod Species | Cephalopod Articles | Lessons | Bookstore | Resources | About TCP | FAQs|
Begging in Cephalopods? - A TCP Exclusive<< Cephalopod Articles | By , Norwegian University of Science and Technology, Trondheim, Norway
Dr. Anderson started his article on begging by giving examples of adult human beggars, young birds, wolf pups, babies, and cats. These examples are pretty uncontroversial. He goes on to suggest that cuttlefish and octopuses have provided the first examples of begging in invertebrates. I am not convinced. The problem lies not with cuttlefish and octopuses being invertebrates (bees and ants have specific behaviours for requesting foodsharing from hivemates, and this behaviour is called begging; therefore the octopuses and cuttlefish would not be the first invertebrates after all), but in the specific behaviours described. I think that if we were to call these behaviours begging, we would also have to call other behaviours begging that do not fit within common understanding of the term.
Dr Anderson cites the following as examples of begging: "we may be familiar with fish begging for food by clustering at the front of a fish tank as we walk by", and further on "A similar simpler sort of begging activity is shown by cuttlefish, which come to the front of their tanks expecting food, much as fish do". I shall start with that and come back to the octopuses later.Stripping the described behaviour down to the bone, the cuttlefish come to the end of the tank where they see something whose appearance is often followed by the appearance of food. Is there similar behaviour that would not be considered begging?There are three relevant behaviours that come from the study of conditioning. Animals commonly approach stimuli that signal a pleasant event, such as the delivery of a reward. They withdraw from stimuli that signal something unpleasant, such as a painful stimulus or the non-delivery of something pleasant. This is called sign tracking. It once featured in a simple reinterpretation of a behaviour that, initially, seemed quite complex. In the 1960s Lilley thought dolphins had their own natural language. A test was devised to find out whether dolphins could talk to each other. Two dolphins were placed in a basin that was divided by an opaque barrier which let through sound. One dolphin, intended to be the sender, was shown one of two signals. One signal meant "press the right paddle", the other meant "press the left paddle". That in itself is not difficult for dolphins to learn. The complication in this experiment was that there were no paddles for the sender dolphin to press. Only the receiver dolphin had response paddles, but could not see the signals. If the receiver made the correct choice, both dolphins got some fish. The idea was that if the dolphins could perform better than at chance level, they could do so only by talking to each other. And the dolphins did do better than chance.
I think it was Robert Boakes who saw a report on this and decided he didn't believe this was proof of dolphin language. He designed an equivalent experiment for pigeons. Two pigeons are kept in two chambers separated by a glass plate. The sender pigeon has in its chamber a little glass disc called a key that can be lit up either red or green. The key is placed so that it cannot be seen by the receiver pigeon in the other chamber. The receiver pigeon has two response keys that are both lit up white when the signal key is either red or green. When the sender sees a green key, both pigeons received food only if the receiver pigeon pecked at the right response key. If the signal key was red, the receiver pigeon had to peck at the left response key. And the pigeons performed better than chance. However, most people are rather more reluctant to credit pigeons with language than dolphins. So how did the pigeons pull off this trick?
Initially, neither pigeon knows what to do. As far as they can tell, they get food half the time when the keys are lit. For the receiver, it makes no difference whether it pecks right or left, it gets food on average half the time at both. Pigeons then tend to develop a response stereotypy. They go for one of the keys, chosen at random. Let's say the receiver always pecks the left key. Once that happens, the sender will get food only after the signal key was red, because those happen to be the occasions when pecking left produces food. Then sign tracking comes into play. The sender will start approaching the key location when the nice red light is shown, and will withdraw from the key when the horrible green light shows. The receiver can't see the signal key, but can see at which end of the chamber the sender pigeon is. And the receiver will notice pretty quickly that pecks at the left key only produce food when the sender pigeon is at one end of its chamber. When the sender is at the other end, there is nothing. Might as well try the right key on those occasions. And it turns out that then pecking at the right key does produce food, so the receiver makes a successful discrimination. As the receiver gets good at this, for the sender the relationship between colour and food delivery breaks down, and so the sender will start approaching the signal key whenever it is lit, regardless of colour. At that point, the receiver then has no reliable indicator for which key it should peck, and can't discriminate any longer. It will again develop a response stereotypy, which then will allow the sender pigeon to signtrack again. The level of success keeps oscillating between chance level and somewhere well above chance. Average across all trials, and the overall level is above chance, yet the pigeons never talk to each other. It is all a consequence of sign tracking, and pigeons' ability to detect positive and negative relationships between events. Coming back to the dolphins, although they could not see each other, the barrier did nothing to their hearing and echolocation.
So sign tracking is simply approaching stimuli that signal pleasant events and withdrawing from stimuli that signal unpleasant events. When the cuttlefish approach Dr Anderson, whose presence is often followed by the appearance of food, do we have to assume it is any more than sign tracking?And if we include sign tracking within the definition of begging, then a pigeon approaching a response key that signals the delivery of food would have to be considered to beg from the response key. Is that really begging?A second relevant behaviour, also shown very nicely by pigeons, is autoshaping. Pigeons will not only approach a signal for reward, they will also peck at it. Furthermore, how they peck depends on what type of reward they can expect. Pigeons peck differently when picking up food than when drinking water. The pecks directed at keys signalling food or water differ accordingly. This is the basis for 'stimulus substitution theory', the idea that in conditioning the signal comes to stand for the signalled event, and that at least some of the same behaviours are directed at the signal as at the event being signalled. That can even interfere with useful behaviour. The classic example is 'the miserly racoon', which was trained to put coins into a piggy bank for a food reward, but ended up holding on to the coins and rubbing them. An experiment with rats showed that misdirected prey handling interferes. The rats were supposed to take ball bearings from one end of a chamber and drop them into a chute at the other end. If they received water as reward, they did fine. If they received food, they chewed the ball bearings, which shared with food the properties of being the right size and solid enough to be chewable. So directing prey handling behaviour at a signal for food is pretty normal behaviour, just an example of autoshaping.
We are not quite at the point where we can explain the behaviour of Anderson's octopus as something other than begging. The behaviour was "Octopuses have learned to go to a particular spot in their tanks to expect food, frequently the front or back of a tank, wherever the human access is. Once on that spot, they will be agitated: they will "pace" back and forth, they will be a bright red color - the color of anger or disturbance. They may do "webovers" (a food gathering technique where they throw the web between their arms over a rock or other potential food hiding area) directed toward the surface of the water - the direction they expect the food to come from. They will turn upside-down, exposing their suckers upward, expecting food to come from above."To explain that, we need to consider a third common behaviour, called goal tracking. Coal tits and great tits (European relatives of chickadees) do not autoshape to a signal for reward because they typically don't sign track in the first place. They have no problem learning that a stimulus shown on a key or a screen signals reward, but what they do is go to the feeder and wait for the reward to appear. That is called goal tracking. Pigeons do it as well once the signal and the reward site are separated widely enough, tits simply do it earlier. What I don't know is what pigeons do once they have been made to goal track. It is conceivable that while waiting for food they then direct the same kinds of peck at the feeder as they would direct at the response key. (I am sure someone must have looked at this, but I don't know the result.)If the prey handling behaviour of autoshaping can go together with goal tracking, we have an explanation for the behaviour of Anderson's panhandler octopus: It sees a signal for food (Dr Anderson), promptly performs some goal tracking by going to the part of the tank where food tends to appear after the signal has been sighted, and directs prey handling behaviour at the exact location where food tends to come from, namely the surface. And if that behaviour is rewarded, the octopus will perform it even more reliably.
Put on the spot by Dr Anderson's article, I can't actually come up with a clear definition that would distinguish the approach to a conditioned signal for reward from begging. The closest I can come to a suggestion for a distinction is to declare begging a social behaviour, and therefore it should only be directed towards stimuli that are also targets for other social behaviour. Rats do make such a distinction: you can use the presence of another rat as a signal for any consequence that you would otherwise signal by a noise or a light, yet the to-be-conditioned rat directs social behaviour towards the rat that serves as signal that would not be directed towards a loudspeaker or a light. Then begging for food should be expected only in species that share food. Young birds begging for food are a good example of that. In some bird species, this behaviour is then carried over into courtship as well. In species where individuals groom or preen each other, they may be expected to beg for that. The only social behaviour in octopuses in the wild that I know of is mating, so I would be surprised to see begging. (As for cat and dog examples, they can be problematic. In the process of domestication, cats and dogs carried behaviours through into adulthood that their non-domesticated relatives show only when young.)
On the cephalopod mailing list, David Scheel responded to my initial scepticism by quoting Webster's dictionary: "to ask earnestly...; to ask for as charity..., as a kindness..., respectfully...". He then argued: "In the context of animal behavior, what does it mean "to ask?" The signal must serve to manipulate the reciever. That is, pardon the imputing of intent, the octopus wants to increase the likelihood that it will be fed."
The criterion that the behaviour should increase the likelihood of getting food would exclude pure sign tracking. Nevertheless, if I arrange things so that a pigeon will only get food if it actually pecks a lighted key, then pecking does increase the likelihood of getting food and the pigeon would still have to be considered to beg from the response key. The criterion of manipulating the receiver does also not capture all the implications of the Webster definition. An anglerfish that induces prey to approach the lure does manipulate the receiver: the deception is definitely a manipulation. But should we say the anglerfish begs from the prey?The Webster definition implies that an autonomous agent is being induced to behave cooperatively or altruistically. I think that does qualify as social behaviour, so perhaps that criterion is not so bad, even though it is not really satisfactory. How do we determine whether a particular behavioural act is part of social behaviour?I am sure there must be people who study begging and who have come up with a workable ethological definition. If any of those people reads this, please get in touch with James Wood and help us out. I have to admit to being reluctant to look it up myself, because I expect it would take a lot of time finding the papers and books that actually contain a definition, then reading enough to judge how well the definition holds up.
I would be quite pleased if Dr Anderson could come up with good counter arguments, because I like to find something unexpected (even if someone else does the finding). Only, I am not yet convinced where begging in cephalopods is concerned.
|Home | What's New? | Cephalopod Species | Cephalopod Articles | Lessons | Resources | About TCP | FAQs | Site Map|
The Cephalopod Page (TCP), © Copyright 1995-2014, was created and is maintained by Dr. James B. Wood, Associate Director of the Waikiki Aquarium which is part of the University of Hawaii. Please see the FAQs page for cephalopod questions, Marine Invertebrates of Bermuda for information on other invertebrates, and MarineBio.org and the Census of Marine Life for general information on marine biology.