1. Introduction

Do cephalopod's have consciousness? Recently, Mather (2008) put forward behavioral evidence for a simple form of consciousness in cephalopod mollusks: octopuses, cuttlefish and squid. Similar brain areas to those in 'higher' vertebrates cannot, of course, be found in cephalopods (though see Wells, 1978; Nixon & Young, 2003 on their brains). But behavioral evidence can give us a basis for asking what an alternate evolutionary path to consciousness might be like, and note that Bekoff, Allen and Burghardt (2002) stress the necessity of looking at the adaptive value of intelligence of the particular kind that is ecologically useful for each animal. This paper takes Mather (2008) forward a step to help answer this question, focusing on content and processes of possible cephalopod consciousness.

Why should cephalopods have cognition and even consciousness? Evolutionary pressure might have come in the development of the coleoid cephalopods (most of the present ones, with the exception of the nautiluses) when they lost the protective molluscan shell. Living in the tropical near-shore, one of the most complex environments in the world, may have pressured the drive to develop intelligence (see Richardson, 2010; Godfrey-Smith, 2000; Kashtan, Noor & Alon, 2007). In addition, Packard (1972) suggested that the coleoids evolved at the same time as the bony fishes and in competition with them for ecological niches. Grasso and Basil (2009), on the basis of newly-discovered learning capacity in nautiluses, believe that the early cephalopods might have been pre-adapted for this ability. While structural parallels with vertebrates are impossible, cephalopods do have big brains (Nixon & Young, 2003). Big brains do not automatically denote complex behavior but they allow for this capacity (Jerison, 2002), and Kortscal e al., (1998) point out the correlation of large brain size with complex marine habitats in fishes. Cephalopods have brain areas (vertical and subfrontal lobes) that are associated with visual learning (Wells, 1978) and spatial memory (Alvez, Boal & Dickel, 2008). Brain size and allocation to different functions varies across the sub-class, as octopuses have 3/5 of their neurons in the arms and much brain area allocated to motor control, and squid have large optic lobes (see Grasso & Basil, 2009). Surprisingly, the deep-sea specialized Vampyroteuthis has a very large vertical lobe (Nixon & Young, 2003), and the function of this allocation is unknown.

2. Content

There are several ways in which simple consciousness in cephalopods does not involve the same content as that of vertebrates. They have little awareness about or behavior addressed to conspecifics. Humphrey (1976) historically felt that social cognition, or the picking up of clues about what other members of your species might be doing, could be the foundation of vertebrate consciousness. Recent research on 'theory of mind' in some mammals and birds has reinforced this, as they appear aware of what other individuals are 'thinking' (Whiten & Byrne, 1988). Cephalopod relationships range from apparently asociality in octopuses (though Abdopus is at least a partial exception, see Huffard, 2008, 2010), through vaguely social cuttlefish and to schooling squid (Boal, 2006). Although Fiorito and Scotto (1992) demonstrated octopuses making the same choice as another that they viewed, this observation of conspecific influence has not been extended. A series of studies by Boal (1996) using chemical cues found no evidence of individual recognition in cuttlefish. They maintained dominance hierarchies, which is common in otherwise solitary animals in laboratory situations, but showed no evidence of recognizing other members of the hierarchy. Octopuses' poor performance on Gallup's (1970) mirror test (Mather & Anderson, 2011) supports a lack of visual recognition of individual conspecifics, including oneself.

Another way in which the cephalopods may differ from vertebrates in content of consciousness is in a lack of conscious monitoring of motor output. For example, with 3/5 of their neurons in their arms, octopuses have much local organization and monitoring. This distribution of the nervous system is probably due to their lack of a fixed skeleton. Extensive neural capacity is needed for coordination of the ensuing hydraulic system of flexible allocation of muscles for support and contraction (Kier & Smith, 1985), as well as the major role of the suckers in exploration of the landscape. Manipulation of arm and sucker musculature to perform tasks is very sophisticated (Grasso, 2008), but we do not know how much information is dealt with at the local level and how much is either controlled by the brain or sent as information to it. However, motor output is not somatotopically represented in the motor centres (Zullo, Sumbre, Agnisola, Flash & Hochner, 2009).

If cephalopods are not social and do not routinely monitor what their arms are doing, why might they have evolved intelligence and consciousness? A clue may be in the demanding environment in which they live, that they must make 'decisions' to find prey and avoid predation in an environment that is not only complex but also constantly changing, and see Godfrey-Smith (2002) for the suggestion that environmental complexity might drive the evolution of intelligence.

One unique cephalopod system whose content could be monitored and output planned is the spectacular display complex (Packard, 1995; Messenger, 2001). Octopuses, cuttlefish and squid use pigmented and reflective color, skin texture changes and posture to match any nearby background, with the exception of color mixtures that these color-blind animals cannot distinguish (Kelman, Osorio & Baddeley, 2008). Yet there has been discussion as to whether this system is open loop, and an example is the counter-shading of cuttlefish, which is reflexive and totally dependent on body position (Ferguson, Messenger & Budelmann, 1994). Is this flexible appearance too complex to monitor, and what is it useful for besides camouflage?

Several different authors have found considerable sophistication in the use of the skin display system to foil predators. The cuttlefish camouflage system reveals sophistication in perceptual assessment (Kelman et al. 2008). Cephalopods commonly flee from potential predators (Mather, 2010) but recruit the skin display system if threat is lesser. In the laboratory, Langridge, Broom & Osorio (2009) found that increased threat from a potentially predatory fish caused cuttlefish to escalate the display system, from a general background resemblance to a clear match, then to the appearance of deimatic eyespots which may make the animal appear threatening. But cuttlefish did not use these displays to non-visual predators such as crabs and dogfish. In the field, Mather (2010) saw a similar differentiation of responses by Caribbean reef squid to approaching fish species. Common but herbivorous parrotfish mostly elicited an 'annoyance' striped Zebra display or deimatic eye spots and were allowed to approach closely, whereas potentially predatory bar jack and yellowtail snapper were sometimes given an eyespot but often a jet-propelled escape response and triggered an escape from further away. Squid could produce eyespots on four corners of the dorsal mantle, and spots were differentially 'aimed' at the approaching fish and not at neighboring squid. Such directionality is also true for the eyespots of the cuttlefish, and not the camouflage coloration (Langridge, 2006). Whether it is conscious or not, much calculation must have programmed responses, based on a lifetime of experience with the particular predator species. In squid, an agonistic display can be addressed on one side and a sexual one on the other at the same time to the appropriate partners (Greibel & Mather, 2003), a 'double signalling' that is unique to cephalopods.

In contrast, Hanlon and colleagues (1999) pursued escaping Hawaiian octopuses across the shallow coral reef and watched them assume an apparently random sequence of display patterns, escape jets and emission of ink clouds. Such a sequence might not be programmed, although changes would break a predator's search image (see Treisman, 1988, for research on humans) and foil pursuit nevertheless. Similarly, there is a series of casual observations on octopuses apparently mimicking the appearance of various poisonous fish and sea snakes. In the best-described study (Hanlon et al., 2010), the long-armed octopus appeared to mimic flatfish swimming and behavior. As mimicry is in the eyes of the beholder, such appearances may not have been attempts to consciously look like other species, and Hanlon et al (2010) point out that this could also serve to place the octopus in an advantageous position to hunt prey in the sand (and see foraging behavior).

A display system which evolved to produce camouflage could be used to communicate with conspecifics, and striking sexual skin displays are found in many cephalopod species. Squid of several species (Hanlon, Smale & Sauer, 1994: Moynihan, 1985; Hanlon & Messenger, 1996; Mather, 2004) have a clear repertoire of major displays such as stripes, bars, pale and dark colors, along with peripheral ones like fin stripes, teardrop around the eyes and arm stripes. Giant cuttlefish 'sneaker' males can display a pattern resembling that of a mature female and fool a guarding male, thereby having an opportunity to mate with the female (Norman, Finn & Tregenza, 2001). Juvenile Caribbean reef squid display a deceptive agonistic Zebra when another larger male is courting a female, and the male attends to them and is temporarily prevented from mating (Mather, 2004). The complexity of the Caribbean reef squid display led Moynihan (1985) to suggest that the skin display system was actually a language, with central skin areas taking the role of nouns and verbs, and peripherals as adjective and adverb ones. However, both Hanlon and Messenger (1996) and Mather (2004) state that the recruitment of peripheral information is basically escalation and that the parallel with language is unrealistic.

With a system of such complexity, it is sometimes difficult to see adaptive use of a specific pattern. Many cephalopods make 'passing clouds' on the skin; Packard and Sanders (1971) first described them for common octopuses, and Mather and Mather (2004) investigated them in more detail in Hawaiian octopuses. A dark patch with contrasting pale margins formed on the dorsal mantle, 'moved' forward across the head and down the outstretched arms with web extended between them, 'flowing' off the edge. The apparent target was a crab that froze, as these clouds formed after an octopus had made an unsuccessful web-over capture attempt. In addition displays were somewhat directional, mostly forward but 'aimed' over a 90 degree range and sometimes unilateral. Note the perceptual manipulation, as the octopus is able to move the cloud by apparent movement without moving itself, thereby preventing retinal slip in its eyes and assuring that its vision is not blurred. Both the timing and the aim suggest that this is by no means an automatic response (and see foraging strategies).

3. Processes

Regardless of the content, any animal will use several processes to manipulate information, and cephalopods demonstrate their intelligence in finding prey and avoiding being consumed themselves. It is difficult to define where processing of information would enter consciousness, certainly it is when choices are made about actions and when an event category or Piagetian schema (see Zacks & Tversky, 2001) is constructed, used and modified, then used again (see Figure 1). Octopuses in particular are learning specialists (Mather, 1995; Hochner, Shomrat & Fiorito, 2006), but much of the research on their learning has been carried out in simplistic and highly controlled situations. West-Eberhardt (2003) reminds us that learning follows the sequence exploration-learning-forgetting- learning, and see Shettleworth (2010) on vertebrate exploration. Few explicit studies have been done on exploration by cephalopods but Boal and colleagues (2000) noted that an octopus placed in an new enclosure spent much time the first few hours exploring, as did cuttlefish (Karson, Boal & Hanlon, 2000). Mather and Anderson (1999) found that octopuses given new items first explored them and then some of the animals later showed motor play.

An external event forces an animal to do what Baars (1997) described as 'shine an attentional spotlight' on the situation. At the opening of the top of their tank, octopuses have a variety of reactions--colour changes, head bobs that are thought to generate motion parallax and construct a better three-dimensional image of the environment, moves or shrinking back from the stimulus (Mather & Anderson, 1999). Similarly, cuttlefish exposed to small local water movements change body pattern, move, orient towards the stimulus, swim away or even burrow in the sand (Komak, Boal, Dickel & Budelmann, 2005). When threatened by an overhead visual stimulus, the cuttlefish showed physiological changes, as well as freezing and hyperinflation of the mantle, which would ready them for a jet-propelled escape response (King & Adamo, 2006). Some of these responses allow the animal to gain more information about the situation and others are preparatory for action. Such information retrieval leads the animal to make 'choices' and 'plans' for behavior in situation such as cuttlefish and octopus maze navigation (Alvez et al. 2008). Feedback about a situation leads to modification of a behavior pattern. Boycott (1954) reported that an octopus stung by a sea anemone on the shell of a hermit crab tried different 'cautious' approaches from different angles, apparently to acquire the prey but avoid the anemone.

One of the ways in which the content of cephalopod consciousness may be similar to that of animals of other phyla is in calculating and using information about their position in the environment (see Shettleworth, 2010). Octopuses shelter in a protective home, and utilization of unusual shelter such as beer bottles (Anderson, Hughes, Mather & Steele, 1999) and coconut shells (Finn, Tregenza & Norman, 2009) may allow them to live in otherwise unsuitable habitat. While they have size and shape preferences, the ideal shelters are not always available. Instead octopuses explore a prospective 'home', clean out a crevice or dig under a rock, pulling and pushing rubble with the arms and blowing sediment out with jets of water from their funnel (Mather, 1994). The octopus may also bring shells or rocks to block the entrance, and there is a significant correlation between the size of the entrance and the number of rocks the octopus has brought to block it. Both the water jetting and the rock stacking meet Beck's (1980) definition of tool use, although tool use does not automatically denote consciouness. Still, the octopus appears to have a 'mental image' of an appropriate home before it begins to dig.

Like many mobile animals (Shettleworth, 2010), cephalopods have navigational ability and, in some circumstances, spatial memory (Alves et al. 2008). In the wild, octopuses move from their central home on foraging paths across the sea bottom (Mather, 1991a; Forsythe & Hanlon, 1997; Leite, Haimovici & Mather, 2009). Chemical cues are not used in trail-following, as octopuses jet through the water and return from a different direction at the end of a hunting trip. They either retrieve stored information about the location of the home via landmarks (visual memory), or remember the turns of their outward track (path navigation) to choose directions. Over several days, juvenile octopuses in Bermuda foraged in different areas of the nearby sea bottom, thus apparently also having procedural memory of the direction in which they had moved in the last few days (Mather, 1991a) and integrating it into a 'win-switch' foraging strategy. Such information is stored in memory as Mather (1991a) recorded detours when octopuses were out foraging; when displaced from their path, they returned directly home from a different direction.

This spatial memory ability has been found in tests in the laboratory which, by definition, must be simpler than the natural environment. Detour experiments with octopuses by Wells (1964) using visual cues seen through glass, had mixed results as octopuses seemed to need to crawl along the wall of the maze, keeping it fixated with one eye. Animals were successful in visually-guided single-turn mazes constructed by Walker et al., (1970), and Boal et al., (2000). Karson et al (2003) extended this paradigm (escape to remembered deep areas when the water level sank) to cuttlefish, and Hvoreckny et al (2007) showed that both species could solve two different mazes that needed different cues when trials were intermixed, proving their ability to make choices of relevant sensory cues.

Sophisticated assessment has allowed researchers to investigate whether cephalopods use a path navigation (monitoring one's turns) or a visual one (memorizing one's position with respect to a landmark). Mather (1991a) trained octopuses to learn to orient to a visual landmark. Karson, Boal & Hanlon (2003) and Alves, Chichery, Boal & Dickel (2007) found that cuttlefish could use either proximal or distal visual cues for such orientation. Those trained with distal cues relied on path navigation and those using proximal cues used both strategies equally often. These results parallel cue usage in vertebrates and insects (Shettleworth, 2010), and suggest a common utilization of similar information in these totally unrelated animals. Interestingly, lesions of the ventral area of the vertical lobe led to deficits in spatial learning (Graindorge, Alves, Darmaillacq, Chichery, Dickel & Belanger, 2006), suggesting this is a parallel area to the mammalian hippocampus. One area of life in which cephalopods, like other animals, may use their simple consciousness is in hunting for and capturing prey. Cephalopods have a wide array of hunting techniques, described by Hanlon & Messenger (1996) as ambushing, stalking, pursuit, speculative hunting and luring. Luring has been suggested when cephalopods wave their arms before an area that potentially contains prey, though this behavior could also be a distraction. More convincing, though not proven by controlled trials, is the behavior of sepiolid squid buried in the sand who stick out a paled arm tip from the sand and wiggle it (Anderson, Mather & Steele, 1999). This resembles the luring of the angler fish and the actions must have some purpose. It is not just that the cephalopods use several foraging strategies, but that individual species and probably individual animals do.

Hanlon and Messenger (1996) present the case of the Caribbean reef squid, which uses ambush from floating seaweed, stalking small fish in the open water and pursuit of larger ones, speculative hunting by touch on the sand bottom to contact shrimp or other buried prey, and possible luring also near the sea bottom. A particularly good example of speculative hunting was observed by the author, when eels were foraging in the coral rubble and squid followed them, ready to snap up escaped prey (and fish do this to octopuses, see Mather, 1992). This activity must have been both learned with a food reward and planned. Long-term learning plays a major role in cuttlefish hunting. Initially newly hatched cuttlefish have a narrow search image for small mysid crustaceans (Wells, 1962; Messenger, 1977; Dickel, Chichery & Chichery, 1997). After a couple of weeks of life, during which the vertical lobe brain area that stores learned information begins to grow, cuttlefish learn to take alternate prey. But visual exposure to the sight of crabs when the cuttlefish are still in their opaque egg cases changes the preference and they accept crabs immediately after hatching (Darmaillacq, Chichery, Shashar & Dickel, 2006). This is a clear parallel to vertebrate social imprinting, early and relatively permanent learning of one's species (Staddon, 1983).

Learning also affects cuttlefish prey capture behavior. Small prey are captured with the extension of the flexible tentacles and larger ones such as crabs are grabbed by all the arms. Naive cuttlefish often attack crabs frontally and consequently are pinched by the claws; they learn to circle around from behind (Wells, 1962) and avoid this defense. Such trial-and-error learning was not improved by social learning when watching another cuttlefish successfully attack crabs (Boal, Wittenberg & Hanlon, 2000), as control cuttlefish exposed only to the odour of crabs were equally successful. The authors suggest that chemical cues might trigger arousal, in this case focusing Baars (2007) attentional spotlight on the situation.

Prey preference leads to acquisition, and octopuses have been described as generalist predators because they capture a wide variety of molluscan and crustacean prey species (Ambrose, 1984; Mather, 1991b; Leite et al, 2009). This lack of choice suggest an unselective generalist predator that is simply capturing any prey item that it encounters as it moves across the sea floor. But a closer look shows that octopuses move to likely locations, probably visually guided, before beginning intense chemotactile search with the arms, a procedure called saltatory search (O'Brien, Evans & Browman, 1989) and found in foraging birds and fish. Such cooperation of information acquisition by different senses, guided by learned information about likely habitat, must have involved storage and retrieval of information and choice of likely areas as well as memorization of areas previously hunted (see navigation). Further, while a population of octopuses in Bonaire selected a wide variety of prey species, individuals had a much narrower range in a varied habitat (Anderson, Wood & Mather, 2008), probably having learned specific prey locations or mastered chosen penetration techniques. The species was a generalist but individual were specialists. Scheel, Lauster & Vincent (2007) evaluate prey availability and capture, and suggest that giant Pacific octopuses in Alaska may be rate-maximizing foragers. Capture of prey does not mean that consumption can begin immediately, especially if the prey is hard-shelled molluscs. Octopuses have three methods of penetration into the shells of clams: pulling the valves apart, drilling a hole into the shell and injecting a poison that weakens the clam's muscles that hold the valves together, or chipping a small piece off the shell margin to allow the same injection. Drill holes are located at different areas of the shell in different prey species, often near the adductor muscle insertions or over the heart of the clam (Nixon & Macconachie, 1988), and the locations are probably learned. Giant Pacific octopus use different techniques with differential success on several clam prey species. Using a schema of procedures, they first try pulling by trial-and-error; if that is not successful, they drill or chip (Anderson & Mather, 2007). Yet each penetration technique also need a different orientation of the clam within the arms of the octopus, out of sight. Pulling is effective when the umbo of the clam is towards the mouth, chipping when the anterior or posterior margin is placed at the same location and drilling when the side of one valve of the clam is pressed towards the salivary papilla near it. This schema must pair proper orientation of the clam with the penetration procedure used to gain access to it. These studies of process bring us back to content. How much does the brain know about what the arms are doing, and how detailed are the output commands? An autotomized arm of a pygmy octopus, separated from the body, can pick up pieces of food and even 'walk' down an outstretched hand. This may be localized processing, and physics can explain the sequence of actions in some arm movements (Guttfreund et al, 1996). But the arms carry out the penetration preparation mentioned above, and those cannot be autonomous choices. Arms 'walk', and the gait can vary depending on direction of travel as well as whether the octopus needs to be camouflaged as drifting seaweed (Huffard, 2006). Arms reach out for prey, and each individual octopus has a 'favourite' arm to extend into a tunnel to obtain a food reward (Byrne, Kuba, Meisel, Greibel & Mather, 2006). Clearly we need to investigate this very different central-peripheral allocation and cooperation to find out how much information is consciously processed.

Are cephalopods conscious? The evidence presented above suggest that they learn, store schemas, attend, recruit information and actions to fit particular choices and attend to specific aspects of a situation when they are relevant (and see Figure 1). Cephalopods studied so far seldom have Fixed Action Patterns. The exception seems to be digging in sand by sepioeids (Mather, 1986; Anderson et al, 2004), using water jets from the funnel to push sand away and throwing sand over their dorsal surface to complete the coverage. Even in this situation, digging is just one choice of several anti-predator actions (Langridge et al, 2009), chosen after chemotactile inspection and with much internal flexibility when investigated in detail (Mather, 1986).

Where might this consciousness develop in thousands of years? Cephalopods are at a metabolic dead end, having made the haemocyanin-based mollusk respiratory system as efficient as possible, with the assistance of three hearts to speed circulation. So they will not likely invade the land. How might their cognitive ability develop? There is no easy passage of information with the overlap of the generations, as in mammals—only octopuses tend young, and only the eggs. Cephalopods are not social, do not cooperate. The skin display system is a magnificent one, its technical capabilities far beyond the sophistication of its use. Squid live in groups and the Caribbean reef squid in small ones, perhaps they will begin to use the display system more nearly to its capacity. Right now the squid have 'nothing important to say' with it. If they become more social, perhaps they will evolve to use it better. Until then, the very solitude of cephalopods blocks their cognitive evolution and further development of consciousness.