« Back to Table of Content || Next »

Chapter II. Evolution of the Human Mind >> Evolutionary pressure drives the development of the hominid’s visual system

Evolutionary pressure drives the development of the hominid’s visual system

Visual identification of still targets

Australopithecines were the first hominid species that left the safety of the treetop canopies and started bipedal walking in savannas. Moving over long distances provided greater access to food. They could have hunted, scavenged nutrient-rich leftovers or gathered edible plants. However, australopithecines walking in savannas had a new problem to confront. A number of predators, including the big cats, hyenas, and the now extinct saber-toothed cats were much stronger than the hominids. The feeble australopithecines (just 1.2 meters (4 feet) tall and weighing only 32 kilograms (71lbs)) had to avoid encounters with the predators or risk being overpowered. Further, the predators were faster than the hominids. Detecting a leopard [5] within several meters would not help avoid an encounter. The hominids had to detect a leopard from several hundred meters away: early detection would let the hominids escape unnoticed or, in case they were hunted, provide enough time advantage to run to a nearby tree. (In fact, there is some archeological evidence that hominids were hunted. Some fossil skulls have paired holes, which fit the gape of saber-toothed cats.)

Defenseless australopithecines (just 1.2 meters (4 feet) tall and weighing only 32 kilograms (71 lbs)) were no match for strong predators abundant in savannas. The big cats and now extinct saber-toothed cats were bigger and faster than the hominids:


Living big cats include the tiger, lion, leopard, and jaguar. These carnivores are characterized by their significant size (up to 250 kilograms (550 lbs), with a body length of up to 3 meters (10 feet)), opportunistic hunting behavior, adaptability to a variety of habitats and ability to move at high speed (up to 60 kilometers per hour (37 miles an hour). The illustration shows a lioness hunting warthogs in the western corridor of the Serengeti.

The saber-toothed cats lived between 35 million and 10,000 years ago. They are most known for having saber looking teeth, which were, in some species, up to 20 centimeters (8 inches) long and extended down from the mouth even when the mouth was closed. They were widespread in Europe, Asia, Africa and North America. The saber-tooth cats were as big as a jaguars (up to 160 kilograms (353 lbs),with a body length of up to 2 meters (7 feet)) and particularly robust front limbs (even compared to modern jaguars). This illustration shows the Smilodon painted by Charles R. Knight. The American Museum of Natural History.

All hyena species are efficient hunters and scavengers. The spotted hyena, shown in this illustration, is the most abundant carnivore on the African continent. Spotted hyenas are powerful pack hunters. Though often labeled incorrectly as a scavenger, the majority of their nourishment is derived from live prey. Hyenas can be as big as 40 kilograms (88 lbs), with a body length of up to 1.2 meters (4 feet).

What sensory organs can detect a leopard from several hundred meters away? Let us consider the olfactory system (sense of smell), the auditory system (sense of sound), and the visual system. The olfactory system detects volatile molecules present in the air. A specific combination of chemicals released by an organism into the air (odor) can supply a unique signature of the odor source. However, the number of volatile molecules released by a leopard into the air is limited. Further, these molecules diffuse randomly in three-dimensional space around the leopard. The concentration of molecules decays with the distance, at a rate proportional to the distance raised to the third power. The chances that a molecule released by a leopard will end up in the nose of a hominid who is located a hundred meters away is close to zero. No chemical sensor can ever reliably detect a leopard from a hundred meters away: neither a sophisticated modern electronic sensor, nor the bloodhounds who are bred for tracking scents. Even at a somewhat closer distance, say, 50 meters (55 yards), a sensory system based on odor detection would be highly unreliable in a savanna where the wind is omnipresent.

Sound can also provide specific information about a predator. Even an immobile leopard generates sound: the heart produces a significant audio signal with every heartbeat. Doctors listen to heart sounds to detect heart problems such as a damaged heart valve. Some animals, like mosquitoes, use heart sounds to detect their prey (mosquitoes also use smell and heat to detect warm-blooded animals). All cars leaving US prisons are scanned for escapees with a device that detects human heart sounds. A driver shuts down the engine, steps out of the car, and the sensitive device detects heart sounds transmitted through the metal structure of the car. Further, sound can be transmitted over great distances. Two people standing on the opposite sides of a lake, a hundred meters away from each other, can talk on a windless day without raising their voice. The sound, reflected by the smooth surface of the water, reaches the ears without much decay. However, the hominids were not on a lake and the leopard would not have been trying to make itself heard. The soft palms of a leopard do not generate much sound and the tall grasses of savanna quickly dissipate sound airwaves, so no heart sounds can be detected from a hundred meters away.

We are left with the sense of vision. In fact, light is transmitted over great distances with little decay. We can observe stars located billions of kilometers away. We can see light reflected from a golf ball flying far in the air. The hominids had to use their sense of vision to detect a leopard from hundreds of meters away. No other sense could have provided better protection. The first adaptation of australopithecines was that of bipedalism. By allowing the head to be above the grass at all times, bipedalism permitted continuous surveillance for predators. A four-legged hominid with its head down foraging in savanna would be an easy target for stronger and faster predators.

However, being able to look over the grass provided only part of the solution. The hominids had to perfect their sense of vision to be able to identify a leopard that was standing still and waiting for them. Identification of still visual targets is an extremely demanding task for the brain. In fact, this task is so difficult than many animals cannot detect a still target at all. For example, snakes have serious difficulties detecting completely still objects: a snake’s visual acuity is based primarily on movement. Some animals are in fact using the snake’s inability to detect still targets as a defense mechanism. When a rabbit detects a snake, instead of running away, it freezes instinctively. There is some chance that the snake’s visual system will confuse the immobile rabbit for a lifeless object, and the rabbit may escape safely. [6]

Even for modern humans, the task of identifying a still, camouflaged animal from the background is a daunting one. Consider the photographs below. Notice how difficult it is to distinguish immobile objects from the background.

Do you see a lizard in this picture? Uroplatus sikorae hide themselves on wood.

Do you see an insect here? If alarmed, a Dead leaf mantis (Deroplatys desiccata) lies motionless on the rainforest floor, disappearing among the real dead leaves.

Can you find two birds in this picture? The Tawny Frogmouth blends in with the color and texture of tree bark.

Do you recognize a snake and a rabbit in this screenshot? I could not recognize either one; I had to watch the movie in order to identify the snake and the rabbit. The visual system identified a moving snake momentarily. If you are like me, please watch the complete Video 2.2.

The above examples demonstrate that, neurologically,  it is much easier to detect a moving target than an immobile one. The neuronal system necessary for detection of immobile targets is much more complex. It requires many more neurons, is energetically costly (one needs to maintain more energy-hungry neurons), and takes much longer to develop (it takes a long time for neurons to grow and establish the right connections, resulting in a longer time during which youngsters remain dependent on their parents). Note that it was easier for snakes to develop a secondary heat-based short-range detection system than to improve a visual-based system capable of detecting still live objects.

For a tree-dwelling ancestor of australopithecines (chimpanzee-like animals that lived over five million years ago) seeing a still target was far less important. The ancestors of australopithecines spent most of their time in or near trees, similar to modern day chimpanzees. While high in the treetop canopies, they were safe from most predators. Even on the ground near the trees they were much safer than the hominids in savannas. To attack a tree-dwelling animal, a predator must move itself toward the tree, making itself much easier to detect. In addition, primates did not have to detect a predator from as far away since they could quickly escape into the safety of nearby trees.

When australopithecines moved away from the trees to savannas, the situation reversed. Now the hominids were moving and the predators were standing still against the background. The australopithecines needed to learn to distinguish a still predator from the background or risk being eaten. The discrimination of predators from the background is complicated by camouflage. Animals tend to have skin color that blends into the background.

The process of visual recognition of predators is that of feature extraction and pattern recognition. It is a familiar task to modern day digital signal processing engineers. It is a very complex task and it often takes years to develop an effective computer algorithm.

Identification of still predators from the background put a lot of pressure on the visual system of australopithecines. The visual system must have become better in every aspect, from eye structure to brain organization. Hominids had to become better and faster in feature extraction and in matching visual cues to memory. The visual system must have also become better at scaling and rotating a visual percept in order to match it to memory. Primate studies indicate that memory volume should not have been a problem. A primate’s memory can store nearly unlimited number of targets. Learning how a leopard looks should not have been a problem either. Hominids were likely to be moving in groups. Any hominid who witnessed his relative eaten by a leopard would have remembered how a leopard looks.

Developing a better visual system took a long time and required more energy-hungry hardware: the brain got larger (the brain of a modern human, which claims approximately 2% of body mass, is responsible for 20 to 25% of body energy consumption. For comparison, other vertebrate brains use as little as 2% of the animal's caloric intake). However, a larger brain must have paid off plenty as any hominid not able to detect a still predator from a long distance was simply eaten by the predator. A better visual system must have also benefitted hunting, foraging (“clever foraging hypothesis” Parker, 1977; Striedter, 2005), and social interactions (“social brain hypothesis” Humphrey, 1976; Byrne, 1988).

A hominid's visual system was pressed to recognize predators based on few available visual details

It is one problem to recognize a still predator that merges into the background when a view of the whole body is available. It is a much more difficult task to recognize a still predator hiding in the grass or behind a tree. The visual system may only see an ear or a part of a tail. The rest of the body may be hidden.

A lion hiding in the grass. Only the lion’s ear is visible above the grass.

There are two ways to recognize a partially obstructed object. One of the ways is to complete available visual cues with missing information and then match the completed image to memory. Psychologists call the process of completion of partially obstructed objects visual amodal completion. This process of matching a whole object to memory is usually referred to as holistic. To recognize a leopard behind a tree using the holistic method, the visual system must (1) separate the visual cues, such as a head and a tail, from the background; (2) to complete the image of the body hidden behind the tree using the process of amodal completion; and only after that (3) match the completed visual percept to targets stored in memory.

Another identification method is to match available visual details (an ear or a tail) directly to targets stored in memory, without the intermediate step of amodal completion. To recognize a leopard using the second method, the visual system must be able to match visual cues to a large number of visual details stored in memory. This method also requires that the visual system had, at some prior time, already disassembled a complete object into elements and stored those elements in memory. To match an ear to a leopard, the representation of the ear must be stored in memory and be associated with a leopard. To match a tail to a leopard, the representation of the tail must be stored in memory and be associated with a leopard. Numerous small details such as eyes, ears, tails, palms, and spots on skin, have to be stored in memory. The analysis of a visual scene is then performed serially: the visual system extracts the first meaningful element from the visual field (an ear visible above the grass) and attempts to match it to memory. It then extracts a second meaningful element (a tail) and matches it to memory, and so on until the object is recognized.

There does not seem to be an established scientific term for this method. Accordingly, we have to create our own term to refer to this process. Since the process involves a prior analysis (breaking down) of visual objects into elements in order to store those elements in memory, and visual analysis at the time of recognition (disassembly of the visual field into meaningful elements for matching to memory), let us call the process visual analysis. Note that analysis involves the disassembly of an object into meaningful parts as opposed to synthesis, which involves combining two or more objects together.

Visual identification of partially obstructed objects. Holistic visual recognition employs amodal completion to fill in missing information and then match the completed image to memory. To recognize a leopard behind a tree using the holistic method, the visual system must (1) separate the visual cues, such as a head and a tail, from the background; (2) complete the body hidden behind the tree in the process of amodal completion; and only after that (3) match the completed visual percept to targets stored in memory. In humans, holistic visual identification via amodal completion is usually the function of the right hemisphere (Corballis, 1999). The visual analysis identification method matches available visual details (an ear or a tail) directly to targets stored in memory, without the intermediate step of amodal completion. In humans, visual analysis is usually the function of the left hemisphere.

Modern humans are great at visual analysis. When we look at a leopard we immediately see the small details: eyes, nostrils, ears, neck, etc. Further we can use a single detail to identify an animal. Consider the following photographs: which tail belongs to a wild boar? Which tail belongs to a leopard?

Modern humans can tell a wild boar’s tail from a leopard’s tail from as far as 100 meters (300 feet) away, and can then react accordingly. We can decide to either hunt the pig or we can decide to run away from the leopard.

Both modern humans and non-human primates are capable of both visual amodal completion and visual analysis (for review see Vallortigara, 2003). However, modern humans seem to be superior at visual analysis and non-human primates seem to be superior at holistic visual recognition. It appears that visual analysis and identification of visual targets based on small details was acquired by hominids at the expense of holistic recognition. Recall the short-term memory experiments conducted by Dr. Matsuzawa. In these experiments five numbers were flashed on the screen for less than a second. After this time, the numbers were replaced by white squares. Subjects had to remember the position of the numbers and touch the white squares in the order corresponding to that of ascending numbers. When the numbers were displayed for about 0.7 seconds, the chimp Ayumu and the college students were both able to correctly remember about 80% of the numbers. However, when the numbers were displayed for just 0.2s, the chimp scored about 80%, while the best humans scored 40%. (Watch the fascinating footage of Dr.Matsuzawa’s experiment: Videos 1.10 A to D.)

From this experiment we are forced to conclude that chimps are better than the humans at remembering a whole image. It is likely that humans perform worse than chimpanzees in this experiment because the human mind automatically rushes to perform visual analysis of the screen followed by matching the numbers on the screen to their representations in memory. In other words, humans spend all their brain resources on visual analysis and matching to memory. The chimpanzee, on the other hand, does not seem to bother with immediate visual analysis. Rather the chimp prefers to remember the whole image and to identify the number locations from memory at a later time.

It seems that this difference in visual strategies reflects the difference in evolutionary history. Hominids, exposed in savannas to stronger and faster predators, developed brain structures responsible for visual analysis, while chimpanzees stayed in the safety of treetop canopies and did not experience a significant evolutionary pressure that could have rewired their brain.

There is strong experimental evidence that in the majority of right-handed modern humans, visual analysis is primarily a function of the left hemisphere and recognition of whole objects via amodal completion is primarily a function of the right hemisphere (Corballis, 1999; these topics are discussed in detail in Chapter III). Thus, it appears that evolution developed the hominid’s brain asymmetrically. The visual recognition in the right hemisphere remained primarily holistic (with all visual cues considered in parallel) similar to that in non-human primates, while the left hemisphere was “trained” in visual analysis (that is, in a sequential analysis of visual cues) and recognition of visual targets based on few available details (an ear, a tail, etc). The training must have been unforgiving. Hominids who were better at both visual analysis and matching visual details to the library of images stored in long-term memory had a better chance of survival. Any hominid not capable of identifying a predator based on the few details available was eaten up by the predator and thus removed from the gene pool.

Australopithecines were probably carrying stones

As soon as my sons started walking upright on their own outside the house (at about 12 months), they spontaneously picked up all kinds of stones and insisted on carrying them home. Nobody asked them to pick up stones. In fact both my wife and I were not fond of them carrying stones home. We may not have expressed our dissatisfaction openly, but it should have been obvious from our facial expressions. However the urge to pick up stones was so strong that both David and Benny carried stones home despite our facial expressions. When their hands could not hold any more stones they gave stones to their parents and picked up even more stones. They continued collecting stones until the age of three. We did not have other children around and both boys grew up at home until they were three years of age. Consequently, they could not have seen other kids picking up stones. There was no TV in our house (there is still no TV) so the kids could not have seen people picking up stones on TV. Finally, there is a twelve-year difference between the two boys so they could not have learned from each other.

I was interested to know whether stone carrying was just a peculiarity of my family, so I asked parents about their children’s stone collecting habits. I was quite surprised to find that a significant majority of boys and girls - over 90 percent of all children - spontaneously picked up stones and sticks and carried them home. From these observations it is reasonable to conclude that picking up and carrying stones is a reflex hard wired into our DNA. There are a number of known primitive reflexes, that is, reflexes that are present at birth and disappear in adults: sucking reflex, tonic neck reflex, Babinski reflex, palmar grasp reflex (stroke an infant’s palm with an object and the fingers will close and grasp the object) and others. These reflexes are hard-wired in the brain from birth. However they are inhibited by the frontal lobes as the child moves through normal child development. I propose that picking up stones and sticks is a primal reflex exhibited when children start walking and inhibited by the frontal lobe at some time during child development (under influence from caregivers).

It is likely that the stone-carrying reflex is evolutionarily old. The development of the reflex probably coincided with bipedalism in australopithecines around five to three million years ago. In fact, australopithecines were the first hominids whose hands were not used for walking or climbing. A stone was an invaluable asset in a savanna. It could assist in protection against some predators and could have been used to improve access to food: breaking nuts and bones (for picking up nutrient-rich brain and bone marrow). Since stones may not have been around when a leopard attacked or when nuts were available, it must have only been natural for australopithecines to carry a stone with them at all times.

Changes in the hominid hand may reflect increased stone handling. In chimpanzees, the thumb is weak and short. The hand is adapted for suspension on tree branches and not adapted for the kind of firm grip that is easily generated by a modern human hand. The hands of Australopithecus afarensis  were still effective for tree-climbing, but the fingers were shorter than those of the chimpanzee and the thumb was longer. The hand of Homo habilis had a longer thumb, with a more human-like grip. Finally, the modern human hand has a longer opposable (or prehensile) thumb, shorter fingers, and some additional muscles, which result in a more flexible, precise, and stronger grip than that of the other hominids. Note that the Neanderthal hand was very similar to the modern human hand. Thus, paleontological evidence is consistent with a continued adaptation of the hominid hand for stone handling.

Comparison of a chimpanzee hand (left) to a modern human hand (right).

Hominids start manufacturing tools

At some point, australopithecines may have discovered that sharp-edged stones work better for protection. In fact, this discovery is not difficult and may have occurred independently many times. If one is attacked by an animal and one has a sharp-edged stone, one would quickly associate the sharp-edged stone with an improved fighting capability. A tribe outfitted with sharp-edged stones could benefit in tasks such as killing smaller animals for food, cutting up carcasses, and protecting their tribe from larger predators.

Australopithecines could have discovered that striking a cobble with a hammerstone splits the cobble into two sharp-edged stones. (After all, even birds are known to drop shells on the pebble beach to extract the meat from inside the shell.) Australopithecines may have also selected naturally sharp stones. However, australopithecines did not manufacture the choppers. Australopithecines lacked some important brain function that would have allowed them to manufacture the choppers. As we have already discussed, the first chopper appears in the archeological record 2.4 million years ago and it is associated with a new species, Homo habilis.

As we discussed earlier, the hominids’ visual system experienced a significant evolutionary pressure. Hominids capable of detecting predators from far away when few visual details were available had a greater chance for escaping from that predator. Identifying a leopard hidden in the grass by just its ear or a tail was undoubtedly more beneficial for survival than not paying attention to those details. The brain learned to disassemble the visual field into details and to match each of those details to previously learned details of similar objects stored in memory in the process of visual analysis. I propose that improvement of the visual system in general and visual analysis in particular enabled Homo habilis to manufacture the first stone choppers.

In fact, chopper manufacturing requires (as a minimum) visual analysis of the cobble into a chopper and a flake. Homo habilis was not simply striking one cobble against another in order to split the stone (as australopithecines probably did), but stroked a cobble with a hammerstone multiple times at the location of choice to make the chopper sharper. Homo habilis was not yet Michelangelo, who famously said, “every block of stone has a statue inside it and it is the task of the sculptor to discover it.” However, Homo habilis was able to see a chopper inside a cobble; he had a mental template of a chopper.

Now we can answer the question of why chimpanzees do not manufacture sharp-edged stone tools. It seems that, in chimpanzees, visual analysis is developed to a lesser degree than in Homo habilis. Chimpanzees cannot voluntarily analyze a cobble into a chopper and flakes. Chimpanzees do not see a chopper inside a cobble. In Chapter I we discussed how primates make tools. We concluded that chimps do not have a mental template of a spear. Rather chimps simply remember to bite on a stick to make it sharp before hunting for bushbabies. If australopithecines made any sharp stones, they simply remembered to strike one stone with anther in order to split the stones. Homo habilis was the first hominid that had a mental template of a stone tool. He was not simply breaking a stone; he acted with a goal in his mind. He was able to see a chopper inside a cobble.

Lack of improvement of stone tools within each hominid species

It is remarkable that within each hominid species, tools did not improve significantly: Homo habilis did not improve Oldowan tools for nearly one million years; Homo erectus did not much improve Acheulean tools for nearly two million years. Even the Neanderthal’s stone tool kit remained relatively unchanged for several hundred thousand years. You cannot explain the lack of development by limitations of hand structure. As we already discussed, Neanderthal hand structure was very similar to that of modern humans.

The stasis in manufactured stone tools within each Homo species can only be explained by the limitations of their brain structure. It appears that each hominid species started at a higher level of cortical development than the previous one, but once that phase had started, further brain development was slow. Evolution developed in steps that significantly improved all cortical organization (in particular, the visual system) and resulted in manufacturing of finer stone tools.

According to this theory, it is not the tools, but the refinement of the visual system, specifically visual analysis, which drove the evolution of hominids. As the visual system improved at performing visual analysis (dissection of visual percept into details) and identification of predators on the basis of these visual details (an eye, an ear, a tail), it was capable of forming a finer mental template. This, in turn, resulted in the hominids’ ability to separate a smaller flake from a flint and, consequently, hominids became capable of manufacturing better stone tools. Homo habilis was only able to break out large flakes from a cobble. Homo ergaster and Homo erectus were able to break smaller flakes. Neanderthals were most likely as good at visual analysis as modern humans. It is the refinement of the visual system that enabled the manufacturing of progressively better tools. In other words, the stone tool record is an indicator of the level of development of the hominids’ visual system.

The first byproduct of visual analysis was the manufacturing of stone tools. The second byproduct was the development of the speech apparatus. The brain, which automatically identified many visual details, needed to be able to generate more words to describe these details.