[ comments ]
How Intelligent Are Cats?
Cat owners often claim that cats are too intelligent to do the sort of tricks that dogs do willingly. Others believe cats are unintelligent because it's harder to train them to do tricks. In this article (on 2 pages) I aim to explain some of these differences and explore feline intelligence and the limitations on feline intelligence. This also means looking at how cats see the world and at some aspects of natural cat behaviour.
Unfortunately for cats, they are often non-consenting participants in surgically intrusive experiments to assess learning and intelligence. Humans seem to feel it necessary to assess the intelligence of animals as a way of reinforcing our own sense of superiority and the cat has been a favourite subject for studying learning and brain function for over a century. Many tests insert electrodes into cats' brains either to monitor brain activity or stimulate certain behaviour; others involve deliberately injuring the brain to see whether learning capacity or intelligence is affected. Most such test subjects are killed and their brains further dissected to look for evidence of brain changes resulting from learning. I personally consider these experiments cruel and gratuitous (their medical benefit to humans is too often dubious) and though some such experiments are referenced here, Messybeast.com does not support this form of experimentation.
In recent years there has been an increase in tests in a more natural home-type environment rather than an artificial laboratory environment. While lab conditions are more easily manipulated, they do not bring out the best in test subjects and give misleading results. Better tests also take into account an animal's innate behaviours and instincts, things which have previously counted against cats in classical laboratory tests. This article also considers some of the anecdotal evidence for intelligence reported by owners, but frequently dismissed by laboratory researchers. Since cats operate in the natural world, it makes sense to observe them in their own environment and not just in highly controlled, artificial laboratory environments.
The Truth About Cats and Dogs
Dogs have been trained to guard/protect, herd, hunt, search/rescue, assist (e.g. guide dogs for the blind) and perform circus tricks, obedience or agility classes. To many, this is a clear sign of their intelligence and the superiority of the canine intellect over feline intelligence. Cats have been trained to perform tricks as seen on films or TV advertisements, but do not have the same repertoire as dogs. This leads to the obvious conclusions that cats are neither intelligent enough nor co-operative enough to be trained.
For example, in experiments where cats and dogs were expected to navigate mazes, most cats performed badly. Dogs soon learned to navigate the maze and reach the reward. Cats sat down and washed. They investigated blind alleys. They did not complete the maze in the allocated time and were therefore judged as "failing the test" or "lackadaisical". Eager-to-please dogs learned that they got a reward for learning the. Cats are not motivated in this way. Being opportunists, investigating every blind alley made sense to the cat - after all, who knows where prey might be hiding in the real world? Sitting down and washing is a displacement activity when a cat is uncertain.
Most of the canine activities cited earlier rely upon manipulating canine social instincts. Dogs live, hunt and play in hierarchical social packs headed by an alpha male and alpha female. They frequently co-operate in raising/guarding the alpha pair's young and co-operate to hunt large prey. Juveniles beg submissively for food from adults. They are eager to please/appease pack-mates in order to remain part of their pack and they demonstrate submissiveness to higher ranking animals. Domestic dogs view humans as dominant pack members so they are eager to please us. In addition, dogs have been selectively bred over hundreds of years to enhance some traits and reduce or eliminate others.
Cats, meanwhile, have a different social structure. Where food is plentiful they are largely solitary although females, usually related ones, may form social groups. Males tend to roam in search of females rather than remain as part of a group. Where food sources are localised (e.g. a rubbish dump) they form colonies but the social structure is more akin to that of lions - groups of females who may co-operatively raise young. Unlike lions, cats do not generally hunt prey larger than themselves and rarely hunt in pairs or groups. Cats are, therefore, independent rather than truly social and have little or no need to co-operate with other cats. Feline co-operation with humans is limited unless it serves the individual cat's interests to perform a task. Whereas dogs have been bred for utility, cats have been bred solely for appearance.
Dogs are largely motivated by the pack-living instinct i.e. they will perform purely for praise and acceptance dished out by the dominant pack member (i.e. the owner or trainer). They will also perform because, in the wild, they risk being driven out of a pack or being demoted to pariah position.
Cats are not motivated by social status factors. To train a cat you must find out what motivates it. Usually this means food, or at least conditioning it that there is the promise of food at the end of the session. Even then, cats are not motivated by food in the same way as dogs - if achieving the food reward is too much hard work, cats frequently cut their losses and go in search of easier "prey". In the wild, it makes no sense for a solo hunter to expend more energy on finding or killing prey than it gets from eating that prey. While dogs will track and pursue prey over long distances and wear down their quarry, cats hunt by waiting in ambush and pursuing prey for short distances only. Starving a cat does not make it easier to train either, cats are better than dogs at ignoring hunger pangs. For young cats, although food is a powerful reward, activities such as manipulation of simple objects such as a ball or scrunched up paper, or the chance to explore an unfamiliar space can be adequate rewards in some tasks. There will always be some cats who not only learn easily, but appear to relish learning, though these are the exception rather than the rule.
Because we judge intelligence by comparing other creatures to ourselves, many popular accounts of cat behaviour describe learning as though cats are mentally defective humans rather than highly specialised carnivores. For example, in 1915 L T Hobhouse (Professor of Sociology at the University of London) wrote:
"I once had a cat which learned to knock at the door by lifting the mat outside and letting it fall. The common account of this proceeding would be that the cat did it in order to get in. It assumes the cats action to be determined by its end. Is the common account wrong? Let us test it by trying explanations found on the more primitive operations of experience. First, then, can we explain the cats action by the association of ideas? The obvious difficulty here is to find the idea or perception which sets the process going. The sight of a door or a mat was not, so far as I am aware, associated in the cats experience with the action which it performed until it had performed it. If there were association, it must be said to work retrogressively. The cat associates the idea of getting in with that of someone coming to the door, and this again with the making of a sound to attract attention, and so forth. Such a series of associations so well adjusted means in reality a set of related elements grasped by the animal and used to determine its action. Ideas of persons, opening doors, attracting attention and so forth, would have no effect unless attached to the existing circumstances. If the cat has such abstract ideas at all, she must have something more - namely, the power of applying them to present perception. The ideas of calling attention and dropping the mat must somehow be brought together. Further, if the process is one of association, it is a strange coincidence that the right associates are chosen. If the cat began on a string of associations starting from the people in the room, she might as easily go on to dwell on the pleasures of getting in, of how she would coax a morsel of fish from one or a saucerful of cream from another, and to spend her time in idle reverie. But she avoids these associations, and selects those suited to her purpose. In short, we find signs on the one hand of the application of ideas, on the other of selection. Both of these features indicate a higher stage than that of sheer association."
Hobhouse interpreted his cats behaviour as having purposeful elements though he offers an alternative "behaviourist" explanation: an association between the motivation and "pleasure" of getting through the door, and the action of lifting and dropping the mat.
Early Stimulus-Response Theories
Early psychologists believed all behaviour resulted from stimulus-response associations. Their theories had no room for thinking, consciousness, instinct, innate behaviours or a predisposition to certain behaviours. At its simplest level, learning involves linking together ("associating") previously unrelated stimuli, or actions and the consequences of those actions. Many invertebrate animals are capable of forming such associations. Early researchers had discovered hard-wired behaviours, but extrapolated that all behaviours were simple stimulus-response reflexes. In 1966, Fernand Mery wrote:
"American neurophysiologists at Yale University are achieving success in a different field. Dr José Delgado installed a complete series of electrodes in the brain of a cat. The operation took place under complete anaesthesia, and when the cat woke up, he knew nothing about what had happened. Experiments did not begin until everything had healed perfectly. It is impossible not to feel for this laboratory cat, but those who were present and took part in the experiment confirm that he made no attempt to escape. He even seemed to appreciate the situation, as if appreciating the interest that was being taken in him. Not knowing anything about the surgical operation to which he had been submitted, he behaved as if he were obeying a simple friendly drill: he became a robot.
Around his neck, one might distinguish a small collar on to which is fixed a receiving set with tiny transmitters, to which are attached neat silver wires, each of which corresponds to a cerebral localisation and disappears into his fur. By this means, whether in the same room or hundreds of miles away, and by a radio-transmitted command, the cat can experience the need to drink (and he has water and milk placed at his disposal), to eat (he can choose whatever he wants), to itch (and can scratch himself as much as he wants). It is even possible, by stimulating such and such a part of the frontal lobes, to provoke in him an overwhelming affection or an aggressive antipathy and, in the very next moment, to reduce these states. The importance of this experiment is not that one can oblige the cat to perform such and such a movement, but one can simply, by passing an electric current, waken in him the desire to act in a determined direction.
At present such experiments towards a better knowledge of feline psychology are not being regularly followed up; though they have been renewed with monkeys and, for some time now, with humans. These same minute electrodes are planted in specifically chosen points which relate to the psychic disorders presented by the subjects. In this way it is possible to make tests whose results are extremely illuminating for psychiatrists. These results are at present being published by the New York Academy of Science. It goes without saying that they may provide us with some frightening perspectives on the human mind."
In the past, psychologists believed all learning to be simple association. The stimulus-response reflex-action theory was also considered true for humans. It is now thought that many mammals are capable of more complex mental processes. Most higher animals have some sort of mental representation of their world, and how the world works, which they consult whenever they have to make a decision. It may never be possible to truly understand how a cat perceives and understands the world. Virtual reality can give us an idea of what the world looks and sounds like to a cat by adjusting the signals which reach our eyes and ears and by filming from cat's-eye level, but however much scientists poke electrodes into the brains of unfortunate felines, they cannot truly get inside their minds. To investigate feline intelligence and learning abilities, we must devise better suited, and more humane, tests. To do that, we must understand how cats have evolved to suit their environment and lifestyle, things that which predispose them to behave in certain ways.
One of the simplest forms of learning is Pavlovian conditioning (Pavlovian Learning). This involves associating a stimulus with an event. One stimulus, called the Unconditioned Stimulus, is normally linked to a particular motivational state and results in an innate reaction called the Unconditioned Response. For example, if the Unconditioned Stimulus is the smell of food and the motivational state is "hunger", then the UCR is drooling! If a Conditioned Stimulus such as a buzzer, occurs just before, or at the same time as, the Unconditioned Stimulus then it results in the Unconditioned Response even on its own. The Unconditioned Response becomes a Conditioned Response and the conditioned subjects drool at the sound of the buzzer.
In a cat's natural environment, an Unconditioned Stimulus might be the pain inflicted by an aggressive tom cat. The Unconditioned Response will probably be flight to avoid a repetition of the pain. In the future, the mere sight of the aggressor (now a Conditioned Stimulus) might result in flight i.e. a Conditioned Response because the cat is motivated to "avoid pain". If the Conditioned Stimulus (the aggressive tom cat) is in the distance the cat is motivated to "avoid detection" and the Conditioned Response is to freeze instead of flee. Pavlovian conditioning forms a link between the original stimulus and the conditioned stimulus, but the actual response depends on the cat's motivational state.
Conditioned learning is complicated by an animal's innate behaviours. Cats' ears are designed to home in on noises like small rustling prey in long grass. In an experiment, arrival of food was signalled by 10 seconds of a clicking sound from a loudspeaker 2 metres away from the food dispenser. The cats ran towards the sound, searched around the loudspeaker or even attacked it. Some ignored the actual food and concentrated their attentions on the loudspeaker. It took hundreds of trials to condition the cats to go to the food dispenser when they heard the clicks. In the same experiment, rats did not investigate the loudspeaker, but quickly associated the sound with the arrival of food. This was not because the cats were stupid. To cats, sound indicates the apparent location of "prey" and they reacted according to their instincts. Highly adapted predators expect to find the prey noises and the prey itself (the food) at the same location. Cats quickly learn when a Conditioned Stimulus is unreliable and they can "un-learn" an unreliable Conditioned Response, ignoring bells, buzzers, clicks or whatever as irrelevant.
Humans are biased in assessing the intelligence of other species, judging them according to their similarity to ourselves. Animals having good eyesight and dextrous hands are consistently rated as more intelligent than animals lacking those features. We are biased towards animals that see, react to and manipulate things in a similar way to ourselves. Animals that learn to do things useful to humans are also rated as more intelligent than less co-operative creatures. This is a shortfall in human worldview, not in animal intelligence.
Animals that rely largely on instinct or highly context-specific learning (i.e. only learns things related to the environment it evolved to live in) can only readapt at a pace determined by evolutionary mechanisms. Those with more extensive learning abilities can alter their behaviour patterns rapidly. Cats also have "ecologically surplus ability" i.e. the capacity to solve problems outside of its specific adaptations to its environmental niche. Ecologically surplus abilities allow animals to cope with rapid or unexpected change in the environment, but are hard to measure. The cat's ecologically surplus abilities are demonstrated by its ability to move from pampered pet to feral feline and back again, within a very few generations, or even within the lifetime of a single cat.
Humans often define intelligence as IQ. This is misleading because there are different scoring systems for IQ and it is possible to learn how to perform well at IQ tests. There are also intelligent people who don't perform well at IQ tests because the tests are biased to certain types of intelligence (e.g. logical reasoning) and are culturally skewed. Other tests include the ability to learn and remember. Is the ability to learn by rote a sign of intelligence? If so, any avian mimic is intelligent. Intelligence comprises many things - the ability to understand and utilise one's environment; the ability learn and remember facts (store knowledge); the ability to link facts; the ability to apply knowledge and to adapt it to new situations; the ability to override or adapt an instinctive response.
A cat or dog does not need to learn nuclear physics or understand Shakespeare in order to survive. Animal intelligence is linked to the animal's natural environment and its survival needs. To measure its intelligence we must adapt our perception of intelligence to its world-view and formulate tests appropriately. If the test relies on learning, we must find out what motivates a dog or a cat to learn or to perform Different animals' ecology means different motivating factors We need tests which apply to the animal's physical and behavioural traits and constraints, not to our own constraints. We also need some way to compare their very different behaviours.
Different animals have different innate behaviours. For example, an untrained cat and an untrained Border Collie dog are both presented with a group of ducklings. The dog herds the ducklings and protects them. The cat stalks the ducklings and eats one or more of them. Is the cat unintelligent because it doesn't herd the ducklings? Is the dog unintelligent because it fails to identify those ducklings as prey and it doesn't take advantage of a meal opportunity?
Neither creature is more or less intelligent than the other if judged by this test. Both performed according to their instinct. The dog came from a breed with a strong herding instinct enhance by human selection over generations; it does what comes naturally to Border Collies. The cat does what comes naturally to cats and identifies an easy meal, but fails the "herding" test. The test is either poorly chosen or is biased towards herding dogs; the results are open to interpretation and the conclusions are worthless. Such tests are sometimes used by "researchers" with hidden agendas i.e. those who simply need statistics to "prove" a pet theory or a foregone conclusion.
Finally, humans are very protective of "intelligence". Indications of intelligence in other animals are often termed "cunning" or are written off as "instinct". As a race, we do not like to admit that intelligence is not exclusively a human trait. Much the same has been true in human history where white Europeans regarded non-white humans (so-called "lesser races") as cunning and able to be trained, but not intelligent. Humans, as well as cats, have a degree of hard-wired behaviours. These hard-wired behaviours allow us to do routine tasks on autopilot and free up more of the brain for solving other challenges.
Horses for Courses - and Tests for Species
An animals ability to master an experimental task often has less to do with intelligence than with constraints imposed by physical traits and behavioural predispositions. Species differ in how they see or hear the visual or auditory cues they are being taught to respond to, just as a human can't learn to respond to an ultrasonic or ultraviolet cue as these fall outside of our hearing and visual ranges. Animals differ in the type of rewards they are willing to work for. They differ in the things they are wary of, or even frightened of, and which will interfere with learning or will entirely undermine an experiment e.g. a cat will not learn to pick a certain plastic shape if the plastic has an offensive odour.
Animals are also predisposed ("prepared") to learn certain types of associations, and are predisposed not to learn others ("contra-prepared"). It is a matter of how their brain-wiring has evolved, predisposing them to interact with their environment in certain ways. If a test or the type of reward doesn't somehow fit into what a cat is predisposed to doing (e.g. manipulating an existing behavioural trait), then the cat won't do it! When trying to measure the relative intelligence of different species (a behaviour that obsesses the human species), some animals do poorly at learning certain things, but if the experiment is redesigned to better suit a species' behavioural or perceptual traits, and it takes into account what the species is predisposed to doing/not doing, the same animals do much better.
Despite being favourite research subjects for over a century, cats are particularly challenging subjects for intelligence testing. It is hard to get them to show how they learn or what they know, especially in a laboratory setting. While social animals like dogs and horses respond to social rewards and to punishment, these are almost meaningless to cats. Although cats may enjoy being petted, it doesn't have the significance of acceptance by a superior in the same way it does for dogs. They are indifferent to the concept of petting as a reward and withholding petting as a punishment; in fact ignoring a cat can be counter-productive as this is a sign of courtesy in feline terms! Punishing a social animal (by ignoring it, speaking harshly or by physical punishment) equates to social disapproval or exclusion from the social group. Whiles this works for dogs, cats are either non-social or have a loose social structure and respond to the same punishment with the fight or flight reaction. Having evolved to be self-sufficient, they lack the urge to appease social superiors or gain acceptance into a pack or herd - they are more likely to go away for a few hours and wait for the human participant to calm down!.
Dogs, rats and other research subjects will learn specific, focussed tasks to gain a food reward. Cats are self-sufficient, solitary, opportunist hunters and have evolved to cope with periods of hunger because only around one in three hunts result in a meal. In experiments where cats which had not been fed for a whole day were tested for their ability to locate an object hidden behind a screen, the researchers noted that the cats searches were "slow" or "lackadaisical" even though the rewards for finding the object were the cats' favourite food treats. In the wild, cats are opportunists and investigate their territory for places likely to conceal prey so the "lackadaisical" test subjects were less motivated by the food treat than by checking all potential prey hiding-holes.
It is obvious to pet owners, and to naturalists observing feral cats, that cats are innately curious and they can and do learn. In the home or natural wild environment, cats adapt their behaviour and strategies according to circumstances. There are cats that play fetch, open door handles or break into packages every bit as fiendish as laboratory puzzle-boxes. Well-designed experiments that take into account the cats physical abilities/limitations and innate feline behavioural traits show cats to be inquisitive, intelligent and able to learn.
Reflex action and conditioned learning are good for some behaviours, but a different type of learning is needed for more flexible behaviour, one which enables the cat to predict the consequences of its own actions, and modify its actions based on past successes and failures. It is a survival requirement that animals learn that some foods are toxic or taste bad after only one mistake and will then avoid that food. This is known as Instrumental Learning or trial-and-error. In Thorndike's puzzle-boxes, cats first clawed and scratched indiscriminately at the sides of the cage, until accidentally discovering the lever, string etc that let them out. Their later attempts were less random.
Some puzzle-boxes were quite complex. One latch required a simultaneous lift and push, and in other cages two or even three latches had to be opened in the correct sequence. Not all cats mastered these, but some did. The skills were gained gradually and Thorndike concluded "The gradual slope of the time-curve, then, shows the absence of reasoning. They represent the wearing smooth of a path in the brain, not the decisions of a rational consciousness." This is a generalisation as some cats improved abruptly and made no further mistakes even if months elapsed between tests. We describe the abrupt improvement as "the penny has dropped" or "something has clicked". One of my cats, Affy, was almost impossible to litter train despite 18 months of effort. One day she watched another cat using a litter tray and "the penny dropped"; from then on she used the litter tray (she had also learned through observation, something I'll look at later in this article).
In early classical psychological experiments, cats readily learned to escape from "puzzle boxes" by manipulating strings or levers in certain sequences. Having learned one puzzle box, they quickly mastered others as any owner of a feline escape artist will confirm. Though they learned to manipulate levers and strings, they could never learn the secret of getting out of the box when the experimenter opened the door to the box only when the cat scratched or licked himself. If a cat accidentally dislodged the latch with its tail, it did not learn anything about where the latch was or how it opened. Associating an instinctive manipulative action like pawing an object with some external real-world consequence is a natural action that the cats brain is predisposed to learn; its natural (which is why so many cats learn to scoop food from a can using their paw like the Arthurs/Kattomeat cat). Associating an instinctive grooming action like licking or scratching with some external real-world consequence is highly unnatural and cats cannot learn it.
In the wild, the skills most useful for survival are acquired most easily. It is easier to train a cat to obtain a food reward by using a normal part of its behavioural repertoire, such as hooking back a bolt with its paw (the same movement is used to dislodge prey that takes refuge in a crevice), than by an arbitrary but straightforward action, such as pushing an identical bolt inwards. Cats instinctively know to hook things out, not to push things further in! However, cats sometimes look for other solutions: In an experiment carried out by Professor Julius Masserman in America, two cats apparently out-thought the humans. They deliberately jammed the mechanism they were meant to operate each time they wanted food. The cats found that by wedging an electric lever into a corner of their cage, the feeder functioned continuously, dispensing food with no further effort on the part of the cats. Whether the cats discovered this by accident and repeated it, wasn't clear in the 1950s report I had. If it is possible to train a cat to operate a lever, it is certainly possible for a cat to learn how to disable the lever.
Another example of associating a manipulative action with a real-world consequence is when your cat scratches "politely" at a door (or window) to attract your attention so that you open the door for it to go in or out. Having learnt the you will open the door for it on at least some of the occasions, it is much harder for the cat to unlearn the lesson. If you ignore it, it will go away and then try again later, so to train it not to expect the door to open, you have to ignore it consistently. One of my cats, Squeak, learnt that pulling a certain branch and releasing it so it hit the door with a loud thump was even more effective at getting the door opened. Of course, Squeak could not know that my real reason for letting her in was to keep the glass panel intact! Many cats also learn that humans communicate through vocalisation and they modify their natural manipulative action (pawing or clawing) and mew at the door or cupboard instead. In essence they are associating 2 lessons (manipulative action + vocal communication) and modifying their own behaviour to get the desired response from their human. Not just a sign of intelligence, but a case of "who is training whom?"
Now back to the puzzle boxes. To your cat, a cat carrier is a puzzle box. Cats learn which side has the opening and often learn to associate the strap-and-buckle fastening with an exit and butt, paw or bite at the door and/or the fastener. If they loosen it enough to escape, the lesson is quickly learnt, often repeated and quickly applied to other cat carriers - having learnt there is a closure mechanism, the cat learns to look for closure mechanisms on any other carrier you put it in. Some owners claim their cats have "learned" to pee in the corner of a cardboard pet carrier and escape through the resulting papier maché - what started as a nervous accident can quickly become a learned behaviour! The problem is, the cat is probably not peeing in order to open the carrier, it is peeing because it is frightened by the carrier (having learned to associate the carrier with the unwelcome ministrations of the vet) and its escape from soggy cardboard is an accidental consequence. The same nervous cats still pee in plastic carriers even after consistently failing to escape from the carrier. Like licking, peeing is an instinctive behaviour and it is unnatural to associate it with an external real-world consequence.
Such intelligence can also be their undoing. Some cats, such as my own Scrapper (one of felinity's brighter sparks), never grasp that cat flaps can be pushed open in both directions; having learnt to push from one side to get out, they awkwardly pull the flap open on the other side when coming in. Cats are also motivated to get into certain types of puzzle box. A food cupboard, a carton or a fridge door is also a puzzle box and the cat soon learns which edge of the door to pull at in order to open it. One enterprising Siamese cat learned to bite a hole in a milk carton, as far down the carton as possible, to get the maximum amount of milk out of it!
Cats view their owners as equals and when a cat tries to please you it does so on its terms, not yours. Cats are also adept at manipulating their owners; those whose cats enjoy playing fetching games might reflect on who taught whom the game. In all likelihood, the cat initiated the retrieving game and trained the owner to throw the object. One of my first cats, Scrapper, regularly retrieved his favourite wand-type toy from a bookshelf and brought it to me - but only when Scrapper wanted a game.
The following series of photos are from psychological testing of cats at brooklynCollege in the early 1940s. The show cats learning to open the puzzle box to get a food reward. In one experiment, 2 cats co-opearted to haul the food towards them. In another, the cats competed to get to the food before the other. And finally, a kitten learns to navigate a maze.
How Cats See the World
How intelligence is expressed is largely determined by how the sense organs and motor abilities (e.g. whether it can manipulate objects) operate. Evolution is economical and an animal's brain is "wired" up according to what sensory inputs it can receive and what its limbs are capable of doing. An animal's brain is wired up according to what is important for its survival. If it relies on vision for hunting, the brain areas related to receiving and processing visual stimuli will be well developed. If it relies on smell, the region for processing smell will be well developed. An important sense gets more brain-space at the expense of a less important one.
The neocortex region ("grey matter") of the brain plays a crucial part in learning and is highly specialised according to species. In diurnal humans it contains a large visual area and a large area for fine motor control of our hands. We excel at intelligence tests that require visual abilities and fine manipulation of objects. Cats are crepuscular (active at dusk/dawn) and rely particularly on their hearing, hence a large region of neocortex is devoted to processing sounds. This is enhanced by their highly mobile ears. The importance of hearing is evident in blind cats, many of which can catch prey or chase toys, relying entirely on sound.
Most humans have excellent colour vision, about 120 o of stereoscopic vision (giving good depth perception), relatively good hearing in a limited frequency range (but not mobile ears) and a comparatively poor sense of smell. We find it hard to imagine how other animals with differently tuned senses perceive the world and intelligence tests were geared towards creatures with human-like sensory abilities. Cats perceive the world quite differently.
Like humans, they have forward facing eyes and stereoscopic vision and can judge size, distance and depth; essential for stalking and pouncing on prey. Cats have about 90o to 130 o of stereoscopic vision, depending on breed-specific traits such as face shape. Otherwise, they view the world quite differently. Intrusive studies measuring electrical nerve impulses in cats' brains show their colour perception is very different. Animals with poor colour vision, do poorly at learning tests which require them to distinguish between different coloured objects.
In brief, the human retina (back of the eye) has three types of cone cell (colour receptors) sensitive to red, green and blue. Nerve cells pick up the relative amounts of red, green and blue and our brain translates this into the various colours of the spectrum. We can distinguish around 100 distinct hues. The other type of cell in the retina are rods; these are sensitive to light and dark. Because we evolved for daytime living, we have relatively few rods and hence have poor vision in dim light.
Cats have cones sensitive to green and blue, but few, if any, cones for red. To a cat, red, orange, yellow and green are seen as one colour. Blue and violet are seen as another colour. Other hues are variations on these two colours (much as monochrome photos are different shades of grey). They can tell that a red object is not black, grey or white, but cannot distinguish it from a green object. Cats are more active in dim light where colour vision is less important than good night vision, so much more of the retina is given over to rod cells. They have enough colour vision to help them spot camouflaged predators, but most owners will have noticed how cats often miss toys (or prey) until the object moves. This is because rods are also very good when it comes to detecting movement (the pattern of light and shade changes when something moves). Cats have other adaptations for dim light. Behind the retina is a reflective layer called the tapetum lucidum. This bounces light back through the retinal cells, amplifying available light (like night-sight binoculars). This is what makes cats eyes glow yellow-green in car headlights or flashlit photos.
Cats have different visual acuity (sharpness) to humans. Acuity is linked to the size and structure of the eye. High visual acuity give a sharper image while lower visual acuity gives a grainier image. Humans can pick out very fine patterns of stripes before the image blurs into solid grey. Testing animals' visual acuity involves measuring brain-wave patterns from electrodes implanted into the brain while the animal is shown a striped image. The stripes are continually narrowed until the signal from the animals visual cortex undergoes a characteristic change, showing that it sees a grey image instead of stripes. A less intrusive method involves training the cat to pick a striped card in preference to a solid grey card, the limit of visual acuity is the point where the success rate is 50/50 for picking the right card. Cats visual acuity is between 4 and 10 times worse than humans. In medical terms, cats have 20/80 vision meaning that what a normally sighted human can see well at 80 feet, a cat can only see in as much detail at 20 feet.
Other visual experiments show that cats can distinguish visual "textures", for example they can distinguish a triangle of vertical lines from a background of horizontal lines. This helps explain why zebra have vertical stripes to blend with vertical lines of the background (trees, tall grass) - a horizontally striped zebra would stick out like a sore thumb to a lion! Cats also see "subjective contours". In the diagrams below, when the three-quarter white circles are properly aligned, an optical illusion produces a black square in the middle of them. When they are randomly aligned, there is no square. Cats can discriminate between the visual illusion and the random patterns.
Cats supplement their sense of vision with extremely sensitive sense of touch thanks to their whiskers (vibrissae). It is general belief that the large cheek whiskers gauge the width of a hole so a cat can tell if it is large enough to get through. As well as the prominent cheek whiskers, cats have smaller whiskers on the muzzle, whiskers above the eyes and whiskers on their lower legs. A blind cat can feel its way over and around obstacles with great precision. The large number of nerves devoted to these whiskers occupy a disproportionately large area of the cat's mental map of its own body (much as the nerves devoted to the hands and fingers dominate in humans).
A cat's sense of smell is far better than that of humans, but is far less than that of dogs. It is, however, good enough that smells imperceptible to us can confound experiments using cats. Hidden food is not so hidden if you are a cat and can smell it. Cats can detect food going stale (and refuse to eat it) long before we can. Smell is an important sense in animals that mark their territories with urine or faeces or that recognise places and individuals by smell.
Cats have excellent hearing and can hear sounds up to about 60,000 Hz while humans (with a few unusual exceptions) can only hear up to bout 20,000 Hz. This means cats can hear the ultrasonic noises made by rats and mice. In addition, they can pinpoint a sound source to within about 8 o thanks to their swivelling ears.
Cats have relatively intricate brain wiring for control of their paws compared to dogs. They are surprisingly dextrous when seizing and manipulating objects. This is most obvious in polydactyl (extra-toed) cats as these often their paws to grasp objects. Photographs and X-rays of cats paws in action show several methods of "handling" an object: it may be pierced with just the claws, held between a claw and pad of the paws, or sometimes held between the paw pads without the use of the claws at all. Cats have some ability to move the digits (toes) of their paws separately, again this is most evident in polydactyl cats. When a cat reaches out to grab an object, it pre-shapes its grip, much as we do, giving it a much better chance of catching and holding the object. Gripping is therefore not simply a mindless reflex action in response to something touching the paw pad.
Early Learning and Slowing Seniors
Psychologists originally believed that animals like cats and humans are born helpless and dependent and develop the ability to learn later in life. Even "helpless" human babies are learning the physical rules of the world around and their innate language module is acquiring language. Exhaustive developmental studies in kittens have found that cats also have an innate mental ability to learn that is present from the start.
Cat workers often comment that kittens develop a preference for suckling from a particular nipple on their mother. Days old kittens can be trained to preferentially suckle from one of two artificial nipples distinguished by texture, location or smell. Using an artificial mother, consisting of a carpeted surface with two rubber nipples, a 2 day old kitten can learn to distinguish between a nipple that delivers milk, and one that does not, based on its texture alone. Discrimination based on odour is possible just one day later. Kittens in pooled litters can also discriminate between its own mother and other lactating females if it is in a pooled litter and between its mother and an artificial nipple.
Despite their mother's protectiveness, kittens have to learn quickly. Orientation develops in the first week. For the first few days, if a kitten is removed from the nest it simply crawls in circles wherever it is. Six day old kittens (i.e. eyes not yet opened) can orient themselves towards the nest in response to the smell of their mother or littermates. By the end of their first week, they have learnt to distinguish by scent the home region of their cage or pen from other parts of the cage. At 2 weeks old, they can orient themselves over a distance of about 3 metres and they begin to explore. Visual cues take over from scent cues at around 3 - 4 weeks. The innate behaviours displayed by kittens are based on inherited patterns, but these behaviours are modified, supplemented and altered, in both the long and short term, by learning.
What determines learning ability is not so much innate brainpower as behavioural development i.e. the ability to take in and process information so it does something useful in the real world. Right from birth, animals, are predisposed to find certain things and certain associations important. They are motivated to explore and learn these important things (or at the very least not to shun those things, even if the exploration stage doesn't happen until they are more mature). Early experiences interact with natural instincts and shape the ability to learn later on. Cats also have different personality types which both affect their willingness to learn and which are affected by early experiences in life. Kittens brought up with other animals, a vacuum cleaner, plenty of people and other household objects are more confident in later life than kittens brought up in a quiet home with only one person.
Just as you can't teach an old dog new tricks, elderly cats are less able to learn. Many geriatric cats suffer a "cognitive dysfunction" syndrome similar to Alzheimers disease and often referred to as feline senility. They are easily disoriented, forgetful, they show compulsive behaviours (one of my senile cats had to be confined because she compulsively walked in a more-or-less straight line until she grew tired and simply sat down), sleep erratically, may forget their litter-training or become incontinent. On a molecular level feline senility resembles Alzheimer's: plaques of a chemical called beta-amyloid appear in the brain. This interferes with the normal action of neurotransmitters (brain chemicals that relay nerve signals) and is also toxic to nerve cells so that nerves are killed off. Even those cats which don't become senile become slower to learn new things.
Studies have found that cats over the age of 10 years are often incapable of learning the basic Pavlovian associations that young cats learn easily. Pavlovian associations are named after the famous Pavlov's dogs experiment where dogs learned to associate a ringing bell with getting a meal and automatically salivated when the bell was rung. Though the older cats were awake and fully alert and their perceptual nerves were supplying the right inputs to their brains, their brains didn't process things as efficiently compared to younger cats.
There is a link between learning, brain activity and sleeping. Researchers have found that different patches of the brain can be in different sleep states at the same time. Sleep regulatory biochemicals build up in the brain during wakefulness and help trigger the transition into sleep. They build up faster in parts of the brain that are most active during wakefulness. The harder a brain region works during the day, perhaps learning a task, the harder that brain region has to sleep at night. Cats that are kept in the dark during wakeful hours have to rely heavily on their whiskers to find their way around; they have unusually shallow non-REM sleep in the visual cortex, but much deeper non-REM sleep in the part of the cortex dealing with touch.
Self-Centred Mental Maps
Some of the apparently stupid things that cats do can be explained by how they mentally map out their world. A cat's world is three-dimensional (includes shelves, tree branches) and is partly mapped by smells which represent territorial boundaries or signposts. The apparently circuitous route a cat might take to get from A to B is not due to stupidity; it is due to the cat avoiding other cats' territories or stopping to check out (or deposit) scents which announce its presence, age, health and breeding status to other cats. These are things to be taken into account when understanding how cats map out their world.
The simplest type of orientation relies on directly seeing the goal, or a step-by-step route based on landmarks ("turn left at the tree, turn right at the fence" etc). Simple orientation systems are error-prone - if a landmark is removed, the is animal immediately lost; something owners of blind cats are well aware of (although blind cats will attempt to find another landmark so they can reorient themselves). Cats use a mix of these methods and construct mental maps of their surroundings, the more thoroughly they have explored, the better their mental map. Cats can also construct mental maps based on a brief view of relevant features, but these are not remembered for more than a few minutes. Mental maps allow cats to take short cuts, cutting across fields instead of sticking to the edges. If given a choice, cats opt for the shortest route to an out-of-sight goal. If there are several equally short routes, they tend to use the one that starts off by pointing in the direction of the goal - something we ourselves do. Minimising the number of twists and turns in a route affects the choice, but is less important than distance and initial direction. When it comes to finding its way around, a cat learns best by doing, not just by seeing.
French comparative psychologists, influenced by the theories of the developmental psychologist Jean Piaget, are interested in how (and whether) various species develop "object permanence." Piaget noted that human infants go through various stages of understanding the physical laws of the world. At first, they lose interest when a toy is hidden or taken out of sight and they make little effort to search for it. Once it is out of sight, it has ceased to exist. Older infants will search for something that partially or completely disappears but may not understand where to look. If they see someone hide the object behind a screen, they will not know to look behind the screen but may instead look in a place they previously found it. As they grow older, they will know to look behind the screen and at around 18 months of age they can follow a series of "invisible displacements": Invisible displacements are when someone hides the ball in a cup, takes the cup behind the screen and takes the ball out of it, then takes the cup back to the infant and shows that it is empty. The infant reasons that the ball is behind the screen. Piaget termed this "Stage 6 object permanence".
Object permanence is a useful skill for animals that need to be aware of the most likely location of prey that has gone to ground. If prey becomes temporarily invisible, a cat first searches for it under or behind the place where it disappeared, but if this is unsuccessful it starts searching the nearest available cover. Cats familiar with their territories know and search the most likely hiding places. Cats sometimes appear unable to solve simple invisible displacement using hidden toys because the apparatus used to hide the toy is equally interesting to the cat! Even though it knows the toy is under a cloth, many cats will play with the cloth (regarded as a new and therefore more interesting toy) rather than hunt the hidden toy. If you roll a ball under a floor-length drape, many cats get distracted and end up playing with the moving drape because it is a new game.
Early experiments suggested cats never reach Stage 6 object permanence. Owners often disputed this finding, based on games with cat toys being lost, hidden or retrieved behind sofas! More recent and better designed studies show that they do reach Stage 6. The cats were tested in their familiar home surroundings and the screens were left around for a week in advance so the cat got used to them and also so they learnt there were no toys hidden behind them. The cats were first taught that whenever they touched their noses to a particular toy they got a food reward. For the actual test, a cat was lightly restrained by its owner and two screens were positioned in front of it. In full view of the cat, the experimenter put the toy in a cup, secretly removed the toy behind one of the screens, and then placed the empty cup in front of the cat. The at was released and, in nearly every trial, went straight behind the screen where the toy had been hidden. The screens were moved from trial to trial and were replaced with new screens of a different appearance, but the cats still got the right answer, proving that they had not just learned a "local rule" but had generalised the solution. Objects do not simply cease to exist and if the object was in the cup before it went behind the screen, but was not in the cup when it emerged again, then the object must logically be behind the screen.
In another test, a cat watched food being hidden in a cup, and the cup was then hidden in turn under three covers, after which the empty cup was shown to the cat. To eliminate scents, the food was not actually deposited under the last cover, but was palmed by the researcher. In one test as soon as the cup was removed from under the final cover and shown to be empty, the cat hurried to this cover (not to the researchers hand). It persistently pushed back the cover until the place where the food should have been was entirely revealed. Not finding any food, it pawed at the cover and tried to push its face underneath for several more minutes. When confronted by prey that has gone to ground, it pays to be persistent (within reason).
In a more complex series of experiments, all sorts of disorienting visual tricks were played between the time the cats saw a toy hidden behind one of several identical-looking screens and the time they were allowed to search for it. In one test, the toy was first placed behind the rightmost of 3 screens. The cat's view was momentarily blocked and all the screens were slid over to the right by a distance exactly equal to the spacing between them. In another test, the cat looked into the experiment chamber from the doorway and after the toy was hidden, the cat's view was blocked while he entire room (walls and all) was shifted to the right. In spite of these tricks, when the cats were released to look for the toy, they found it by using an absolute sense of position (a course and bearing from its own position) rather than a relative one. They did not look for it behind what was now the rightmost screen, instead they looked behind the screen that now occupied the precise spot in space that the rightmost screen had previously occupied when the toy was hidden. A cat's sense of space is "egocentric" - they remembered where the toy was placed relative to their own fixed position in space, and not by the toy's position relative to a landmark.
When the experiment was set up to make egocentric spatial reasoning impossible, the cats were forced to orient themselves using landmarks. From a central doorway, the cats observed the toy being hidden. However, they could only enter the room by taking a detour through an L-shaped tunnel, entering the room through a door to either the left or the right of the one they had watched from. Unable to use an absolute sense of position. These cats successfully located the toy using landmarks. If the egocentric cues and the landmark cues conflicted, the cats trusted to their own cat-centred co-ordinates.
Cats form a mental map of their environment, but instead of mapping landmarks ("the church is 300 ft to the left of the shop, the shop is a mile north of the farm") a cat's mental map has the cat in the middle and everything else is relative to the cat's position. This explains why cats do some apparently stupid things, such as failing to cotton on to a moved litter tray even if they watched you move the litter tray a moment ago, and why they are such creatures of habit. It takes time to adjust the egocentric co-ordinate system, hence moving the litter tray should be done by shifting it a foot or so each day and moving the feeding station should be done by establishing two feeding stations and only removing the old one when the cat has got a co-ordinates fix on the new one. It's not that cat's are stupid, it's just that their internal maps is different from ours.
The Feline Time-Space Continuum
Many species have specialised modules of the brain for certain tasks. Species which cache nuts and seeds for the winter have a phenomenal spatial memory (and a correspondingly large hippocampus region of the brain). London taxi-drivers who have to remember lots of routes and street locations also tend to have a relatively large hippocampus. Humans have a highly developed "language module" and human infants can acquire language, complete with rules of grammar, just by listening to it. Border Collies instinctively herd things. Experiments to assess animal intelligence often overlook or dismiss them innate or instinctive skills as being unrelated to intelligence. Instinctive skills may still require a huge amount of brainpower; by hardwiring them as instincts, the animal is spared the overhead of having to learn them from scratch, but it must still hone these skills.
Cats instinctively hunt things. Even if they don't hunt prey, they show hunting behaviour when playing with toys, playing with other cats or playing with owners. Hunting involves knowing where to find prey, following the motion of fast-moving prey and co-ordinating the motion of paws and jaws to seize the prey. As kittens, a lot of feline play is geared to honing these instincts. The basic hunting skills are hard-wired into the cat's brain. Even if a cat has never hunted, the pounce-and-bite behaviour can be triggered by stimulating the appropriate part of the brain with an electrode inserted into it (like the poor feline "robots" described by Fernand Mery). The behaviour is automatic and even if the cat is not hungry it will still react to the stimulus whether it is an electrode or the sight and sound of prey. In the wild, a cat cannot afford to pass up a chance to catch a meal (in the wild, a cat is rarely so well fed it can't manage another meal!).
Many owners have seen their cats watching nature programs on TV. Most cats quickly put the TV into the same mental category as a window - they can see and hear the animals, but can't reach them. After one or two investigations behind the TV or the speakers, they learn that the animals stay inside the box. After that they don't bother checking for escaped TV animals again, or at least don't expect to find anything if they do check - when you are a cat, it can't hurt to be absolutely sure there isn't a snack-sized wildebeest behind the TV! The interesting thing is cats recognise TV images of wildebeest as being potential prey. The secret is they recognise how animals move.
Cats can tell the difference between the motion of a living thing such as a mouse or a TV image of a wildebeest and the motion of an inanimate object such as a blown leaf or a rolled ball. In one experiment, cats were shown moving images on two computer screens. One image contained 14 dots that represented the outline of a walking or running cat. The other contained 14 randomly moving dots. The cats consistently distinguished between the interesting animal motion dots (animals = food potential) and the less interesting random dots. However, if the animal motion computer screen was turned upside down, the cats could no longer distinguish it from the random motion screen. To a cat, animals running upside down make no logical sense. Modern AI programmes have problems recognising animal motion dots even when they are the right way up.
A famous specialised feline instinct is that of landing on all four feet, known as the self-righting reaction. In experiments, young kittens were dropped 40 cm (16 inches) onto a cushioned surface. At 4 weeks old, they lacked the ability to right themselves. Between 4 and 6 weeks old their self-righting ability developed and improved until at 6 weeks old they consistently landed on their feet. Though the instinct is hard-wired into the cat's brain, it has to be honed and the usual time for honing it is when curious kittens fall out of trees or off of furniture. In cats with normal motor abilities, but certain types of brain damage, the self-righting reaction is lost and seemingly cannot be learnt from scratch (noted through observations of pet cats). Adult cats have been trained to demonstrate their self-righting ability for time-lapse photography. Having worked out the distance they are falling (the same every time), some cats became "lazy" and left self-righting to the last moment! These "lazy" cats demonstrate that cats have a remarkable sense of time as we will see later on.
Some animals, such as the seed-hiding birds and fruit-eating monkeys, have excellent spatial intelligence. They can find their way to a series of fixed sites (caches or trees) using the safest or most efficient routes. In addition, some animals optimise their routes so they visit the richest food sites first. Cats are opportunist hunters and do not follow such carefully planned routes. They probably don't decide in advance what sort of prey they are going to hunt. Of those cats that rely on hunting, for example farm or feral cats, they spend only a few hours each day hunting and the typical hunting trip is less than 30 minutes. This was reflected in laboratory experiments which show that learning certain kinds of spatial relationships does not come naturally to most cats due to the egocentric mental maps (and the use of scent markers on vertical surfaces).
Though complex spatial relationships may not come naturally to cats, remembering a simple location does. Having learned that prey (or cat food) is usually to be found in a particular location, cats will return to the location. Moreover, they associate the availability of food with a time of day or time interval: cats are very good at time calculations as the owners of "furry feline alarm clocks with no snooze button" can confirm. Cats appear to calculate how much time to invest in hunting and can discriminate time intervals with an impressive degree of precision. For a cat, the time interval between hunting trips and the energy expended on a hunting trip are more important than the spatial relationship between areas where food is obtained.
Cats can tell the difference between a sound that lasts 4 seconds from one that lasts 5 seconds and can learn to delay their response to a stimulus by several seconds, down to an accuracy of one second. This means they have an internal clock, with a one second accuracy, that can be used to time both external and internal events.
In one experiment, cats were placed in cages for either 5 seconds or 20 seconds. When released, they were rewarded with a food treat that would always be hidden in the left-hand feeder if they had been in the cage for 20seconds and in the right-hand feeder if they had been in the cage for 5 seconds. If the cat went to the wrong feeder, it was counted as an error. After training 14 cats, using 400 - 1000 repetitions of the drill each (depending on the cat), all 14 cats could pick the correct feeder more than 80% of the time. The researchers then shortened the 20 second trials to see if the cats could still tell the difference between a long wait and a short wait. 7 of the cats could discriminate a 5 second interval from an 8 second interval.
In another experiment cats were trained to press a bar a number of times to open a food tray; having gained access, they could eat as much as they wanted at that sitting. At first it took 40 presses to gain access to the food. As the number of bar presses required for the food tray to open was increased (up to 2560), the cats responded by eating fewer meals each day, but eating more at each sitting. The cats were not counting the presses (we'll look at number sense later on), they simply continued pressing the bar until the food tray opened. For a cat to press a bar 2560 times shows a remarkable level of patience and persistence. The trade off was to expend less effort but more often, or expend more effort but less frequently. Researchers then varied the number of bar presses from one meal to the next, the cats calculated the average "price per meal". They amount they ate at a given meal was related to the average number of times they had pressed the bar in the course of a whole day or over a period of several days, not to the number of times they had pressed it for that particular meal.
According to psychology lecturer Britta Osthaus at the University of Exeter, cats do not understand cause and effect. She expert attached fish and biscuit treats to one end of a piece of string and placed these under a plastic screen to see if the cats were able to work out that pulling on the string would pull the treat closer. The cats were tested using a single baited string, two parallel strings where only one was baited, and two crossed strings where only one was baited. All cats succeeded at pulling a single string to obtain a treat (93% of the time) showing they were able to learn the connection between the string and the treat, but none of the cats consistently chose the correct string when two strings were parallel. When tested with two crossed strings one cat chose the wrong string consistently and all of the others performed at chance level. According to Osthaus, dogs were able to solve the parallel string test, but cats weren't. This test was flawed. Firstly, cats are less food motivated as dogs, and are as likely to be interested in the string as a toy as in achieving a treat. Secondly, the comparison with dogs was also incorrect as another paper, co-authored by Osthaus - if the strings were placed at an angle or were crossed, the dogs tended to paw or mouth at the location closest in line with the treat. In other words, both cats and dogs understood the means-end connections involving strings, but they were both unable to understand crossed strings - something very different from failing to understand cause and effect. Dogs evolved as pack hunters that may select a single animals from a herd - not dissimilar from selecting a string that will give a food-reward. Cats evolved to stalk single prey rather than making choices in that way. If a cat has previously found a mouse at a certain mouse-hole, it makes sense for the cat to check that "empty" mouse-hole again as other mice may be there. In this way of thinking, it makes sense for the cat to check the empty string that previously had a food payoff. Dogs make choices when pursuing prey, cats investigate all available bolt-holes. If you design a test that favours the dog's natural behaviour and view of the world then the dog will appear to perform better. Pet cats have learnt how to open doors using door-knobs and experimental cats have learnt to dispense food using a lever; both instances of cause and effect.
When cats do deign to co-operate on traditional animal intelligence and learning tests, they perform quite well. As cat owners well know, cats clearly indicate when they are bored of the "game", which means a lot of patience is needed on the part of the testers. Cats do not like frustration and will often give up or select random answers when faced with situations where there is no clear path to a pay-off. In the wild, a cat frustrated by elusive prey will eventually go and hunt something easier instead; it makes a trade-off between time and energy spent and the likelihood of a worthwhile meal. In intelligence testing, cats "learn to learn" when rewarded for their efforts, but they will learn to "not bother learning" when faced with problems with unclear goals and no guarantee of a reward.
L.T. Hobhouse's experiments consisted of simple puzzles that his animals had to solve to get a food reward, though he noted that the cat's innate nature made it a difficult subject. "My first experiment was with my cat Tim, a small black tom, rather more than a year old. .... Tim is a sociable creature, who follows his friends about in the half dog-like way that some cats have, but as a psychologist he has two great defects. His attention is of the most fickle order, and what is even worse, he gets his meals at the most irregular times, and by methods known only to himself. It is therefore impossible to say beforehand whether he will take any sustained interest in the proceedings at all."
Here is one of Hobhouses experiments: "A piece of meat was placed on a card to which a string was tied, and then placed on a shelf beyond reach of the animal with the string dangling down . . . . I first tried this with Tim, thinking that a young cat would very likely pull the string in play. I was surprised to find that he took no notice of it. I showed him seven times, pulling the string down before his eyes, and letting him get the meat. Neither this, nor a series of trials in which the card was placed on the table barely out of the cats reach, had the slightest effect. The kitten once grabbed the string as I was arranging the card, probably in play, and brought the card down without the meat. For the rest, he either made no attempt at all, or tried to claw at the meat directly. About a fortnight afterwards I began a long series of trials in which the string was tied to a chair leg to make it more conspicuous. . . . Fourteen trials gave no result. Next day, eight trials passed without result, but at the ninth, the cat bit slightly at the string close by my fingers as I adjusted it, and as soon as I had got it right, pawed the string down. The biting was doubtless due to the string being slightly smeared with fish, but the effect was apparently to call the cats attention to the string for the first time in all this long series. It is clear that, in pawing it, his aim was to get the fish on the table. If he had merely been attracted by the smear on the string, he would have used his mouth. At the next trial, he sat still for a while, and then pawed the string again. At the next, he took to washing himself, and I gave up for a time; but on replacing the string I saw him watching me, and he pulled it down at once. In the next trial he did the same. Next day he appeared to have forgotten, but walked under the string and knocked it down with his tail. At the second trial, he slightly brushed against the string, but walked away. I had to rearrange it. He watched me doing so, and pawed it down at once. He then pulled it five times running without hesitation."
The cat, it seemed, treated the experiment as a game (although Hobhouse did not actually say this). There are reasons for its repeated failure to understand what was expected of it. It might have had difficulty recognizing the relevance of the thin string, particularly as cats are long-sighted and it might not have been able to see the string properly. Alternatively, the first time it pulled the card down there was no reward and the cat immediately lost interest; it was much more interested in the smell of fish later on. On a later occasion, the reward of fish came at the first attempt and the cat was then quick to learn the trick. Hobhouse had discovered how easily cats are demotivated.
In one set of experiments cats are presented with a pair of mismatched wooden figures which might differ in shape, size or colour e.g. a black square to the left of a white circle. The cat chooses one or other object by nosing it and every time he picks, for example, the black square on the left hand side, he is rewarded with food. Once the cat consistently picks the black square, the experimenters randomly switch the black square to the left or right of the white circle. After much patient repetition, the cats get the hang of picking the black square rather than whatever shape is on the left hand side (the "success" criteria is picking the correct shape 80% of the time since most cats occasionally check out the other shape, just in case). Later the white circle might be exchanged for a different shape such as a white triangle, or even a white square, and the cat learns to pick the original black square no matter what the other shape is. Similar object discrimination tasks have been used to assess other aspects of feline intelligence, not just whether it can tell the difference between shapes, colours and textures.
Having learnt the correct solution to one such "object discrimination" problem, cats can learn to generalise from the experience. They catch on faster to similar object discrimination problems. To begin with, each new pair of objects requires dozens of repetitions before cats hit the magical 80% mark. After mastering about 60 different object discrimination problems, many cats will hit the 80% mark after only 10 trials. In other words, the cats have learnt that the rules of the game are to work out which of 2 objects results in a reward.
Cats can extrapolate from right answers, but are not so good at extrapolating from wrong answers and end up becoming discouraged, bored and unco-operative if they keep getting a test wrong. If the test cat is lucky enough to get the right answer and its reward on the first try, he masters the problem much faster than if he picks the wrong, unrewarded answer the first try. This is not due to lack of intelligence, but is to do with a hunting animal's innate behaviour. If a mouse is not found at the first location a cat visits, the cat does not automatically visit the second location - cats are opportunist hunters and do not follow fixed search patterns. By contrast, foraging animals visit a fixed set of likely food sources, starting with the most likely food source first.
Cats won't tolerate frustrating situations for long and quickly give up or become indifferent when there is no clear path to a reward. So they have a harder time with a problem where they have to learn to pick an object on a given side, either the left or right, depending on which of two possible pairs of identical objects (e.g. 2 black squares versus 2 white circles) is presented. This problem has no equivalent in the cat's natural world, so they have difficulty learning what is expected of them. Many cats eventually learn to solve "tough" problems like this, but their performance is generally only better than chance. They also have more problems extrapolating from right answers when presented with a new "tough" test.
Cats that are given a mix of simple and tough problems catch on faster to the tough problems than do cats who are given a straight course of nothing but the tough problems. One cat who had only ever been presented with "tough" hard problems, never learnt to master a simple black/white discrimination task despite 600 trials. With no equivalent challenge in nature, cats presented with only "tough" tests become demotivated and appear content to get an occasional handout when they choose the right answer by chance. In certain types of test, intelligent cats are content to underachieve - a problem with the design of the test, not with the cats' intelligence!
[ comments ]