The Semantic Trap
It seems nearly impossible for a writer tackling the consciousness problem to avoid a linguistic trap. Ultimately, when we describe consciousness as an act of ‘observing’ ourselves in the act of observing the environment and our physical states we are faced with a seeming paradox. I am observing myself observing the world, but who is the “I” doing the observing? Is there an “I” observing the world? Is there another “I” observing the first “I”?
Douglas Hofstadter, in his 2007 book, I Am a Strange Loop claimed that these seemingly multiple “Is” were really one and the same observer and that the observer is a self-referential entity that is observing itself observing.
On the other hand, Antonio Damasio, in The Feeling of What Happens has constructed a layered architecture in which a series of “maps”, e.g. cortical and sub-cortical structures that process images act as observers of the layers below (see figure below). That is, information about what is happening in the world and in the body passes upward toward layers that will use this information, along with knowledge encoded in memory and innate dispositions (from lower levels) to decide what actions should be taken. This is basic consciousness, or what Damasio calls “core consciousness.” Much of the functions contributing to core consciousness are performed in the more primitive parts of the lower brain. The five lower maps shown below are formed in various cortical areas in midbrain and paleocortical structures such as the amygdala and hippocampus. The two higher level maps were possible as soon as structures such as the cingulate cortex could encode experiential knowledge and provide comparative analysis between what is happening and what memory predicted would happen.
Figure 1. From Figure 3 in prior posting.
The evolution of the neocortex in mammals added a spectacular capability in forming much more complex maps as well as greatly expanded monitoring of the biographical self. The neocortex is a massive memory system (c.f. Fuster, 1999) that allows an animal to build a historical record of experiences and how those experiences affected the organism. This memory system is actually a learned model of how the world works and how the organism works in that world. In figure 2 I show what I call a “Super-observer model” that is dependent on neocortex sufficient to integrate experiences into the developing model. The new map/model is capable of asking more advanced questions regarding how successful the organism has been over its lifetime at dealing with and adapting to the environment it has been interacting within.
The organism now maintains a more explicit sense of the self. The model that is constructed within the neocortex includes a model of the self. Most of the framework of the model comes from the core self, the brain's inherent and generally fixed operational, basic logistics, and basic tactical models (e.g. the amygdala monitors for threats in the sensory fields). But the neocortex-based model allows for incorporation of learned experience into that framework. For example the amygdala might be leery of any snake-like (slithering) thing that it detects. It would initiate a response in the form, usually, of getting away. Evolution “trained” this structure to recognize and respond to snakes because, over the history of the species and its predecessors, snakes have proven to generally be dangerous (circuits in the amygdala that triggered safety responses were selected for, those that didn't got their possessors bitten!) But with a neocortical model, the organism might learn that not all slithering is done by snakes and so even if the amygdala tries to initiate escape, the early prefrontal cortex could check the model and re-check the source of the trigger and decide to not escape if it isn't warranted. This is how dogs and cats can learn to be friends!
The self model in neocortex is far more malleable than the core model. It can override the core model (as just explained) under appropriate conditions and this has tremendous fitness benefits for animals that explore larger and more complex environments for increased opportunities for food and mates. The core model is course-grained. It treats many different objects as threats or opportunities that really aren't in the larger context. If you responded by running away every time your amygdala detected what it thought was a threat you would miss some opportunities that could contribute to your success.
Figure 2. The expansion of core consciousness depends on a higher-order map that has access to a biographical memory.
Mammals and many birds behave as if they had a sense of self. That is they do not merely react to stimuli in pre-programmed ways all the time (as a reptile generally does). They show the ability to problem-solve in novel situations, such as getting a food morsel out of a bottle.
A sense of self comes from monitoring the impacts that things happening in the environment have on the organism's body. Every action-result pairing has some affective impact, such as when food is successfully obtained the ‘pleasure’ associated with the action to get it has a reinforcing effect on learning that pairing. Rewarding stimuli-action pairs makes those more likely to be chosen in the future. Similarly punished behaviors are made less likely. Damasio (1994) describes what he calls “somatic-markers”, or conditions of the body state concurrent with the mental images being processed, where affective (emotional) states can be associated with newly formed memory traces so that in the future the activation of those traces can be influenced by the emotional content. For example a memory that was formed under negative emotional states may affect a current condition decision if that trace is re-evoked in the present. The possessor will tend to avoid whatever triggered the recall.
The Great Leap Forward
Primates evolved more extensive prefrontal cortical areas that were larger proportionally to the rest of the neocortex (for the size of the brain). The prefrontal cortex sits at the very apex of the behavior decision processing. It has been said to perform the “executive” functions, which includes things like directing attention, calling relevant dormant memories into working memory, even just deciding what needs to be decided. These brain circuits emerged as management specialists to coordinate the rest of the neocortex in their functions. Fuster asserts that they evolved from more primitive motor cortex in the frontal lobe and are, in fact, specialized action (motor) cortex. But the multiple coordinators needed to be coordinated among one another. For example there is a circuit that is responsible for evoking some functions in the cingulate cortex that will make comparisons between items in sensory vs. working memory. Based on the outputs of the cingulate cortex another patch of circuitry decides how to integrate the new sensory information into the exiting model (active in working memory). These circuits need to be coordinated in action and timing in order to produce the proper sequence of processing steps.
The precursors for these circuits already existed in older frontal lobe cortex. The “Super-observer” not only observed what is happening now relative to what has happened in the history of the individual, it also became responsible for considering alternative behavior options in order to add more exploratory possibilities to the repertoire, Rather than be completely dependent on already learned behaviors, mammals evolved increasing ability to try new possibilities. This required a higher order judgment capability than the simpler comparison analysis provided by the older cortices. This too was accomplished in the prefrontal circuits that sat right at the right point of convergence of all the information flows.
The patch of tissue in the prefrontal cortex, right at the very apex, right behind the eyebrows (in human skulls) evolved as the “final” model/map that became responsible for coordinating all of the coordinators (the rest of the prefrontal cortex). To do this the capacity to ask what alternative behaviors might be tried needed to become a more organized and motivated function. It had to coordinate all of the other resources of the brain to formulate plans for future activities. The organism is always driven by biological needs. These are the motivations that drive the organism. The long-range plans (which need not be conscious, ironically) have to support meeting these needs.
Consider, for example, the biological mandate to reproduce. We humans, and presumably all animals, are not consciously aware of a drive to reproduce. Instead there exist subtle and powerful drives to have sex, which provides rewards to the organism (dopamine shots!). But humans construct elaborate models of love and bonding relations that constitute their basic ‘plans’ for ultimately consummating a fertilization event leading to one or more children. All of our conscious thoughts about love and mates, how to choose the ‘right’ one, etc. constitute our plans and whatever they turn out to be (and they can be incredibly variable in content) they will drive us to make choices in daily life that should lead to success in reproduction. That is, if the plans we constructed are realistic. A major problem with the human condition, being subject to the developmental (learning) influences of a complex culture, is that too often, especially in modern life, those plans are essentially unrealistic. An example is the Hollywood version of love and ‘happy ever after.’ Most people (present company not excepted!) have tremendously wrong visions of what marriage will be when they achieve it. This is a major downside to being conscious and non-conscious simultaneously.
Figure 3. Part of the prefrontal cortex adds a new kind of map/model that works on planning longer-term courses of action or “strategies.”
The sense of having a purpose emerges from the construction of a planning model and the on-going monitoring of the execution and success of the plan produces the sense of agency — the self power to exercise control over the self's situation especially by controlling elements of the environment.
Too often the plan simply isn't viable. It does not represent the real world, or the real self, as they actually are. And this can lead to a major disconnect between expectations (to carry out the contents of the plan) and actual results — that sense of power over the environment is thwarted. The frustration that builds up internally as a result of continuous disappointment, whether consciously experienced or not, leads to mental breakdowns sooner or later. Our modern culture seems to be particularly good at influencing the construction of unrealistic plans and unrealistic beliefs about how the world works and who we are as persons.
The smallish patch (designated as Brodmann area 10, BA10, in brain area maps) isn't a homunculus. It does not accomplish all of these functions and be the seat of the memories of the senses of purpose or agency. It is just a super coordinator whose task is to coordinate all of the rest of the brain components to accomplish these results. This model of the brain and the functions of regions fit the hierarchical cybernetic model of operational, logistical, tactical, and strategic management (see my post: The Science of Systems 7. The control is over the vast memory stores in that hierarchical fashion. BA10 coordinates existing prefrontal areas for long-range plans and monitoring. They, in turn, coordinate multiple other areas in association and pre-motor cortices.
Human consciousness, at its best, involves a brain actively observing its own working, which is primarily the organization of memories on the basis of prior experience and current situations but mediated by memories of the future (plans). But this level of consciousness is not always active. Most of our lives are governed by current perceptions, especially when we are concentrating on accomplishing specific tasks. We only enter this consciousness of consciousness when demands from the external world are not keeping us focused and responding.
Over the last one hundred thousand years (or more) human evolution has seen a tremendous jump in the size and connectivity of BA10. One of the most important effects that may have been a result of this is the great increase in level of judgment, that is judgments over much more complex and longer-time scale issues. Additionally human empathy and altruism have increased in concert with the expansion of BA10 (and judgment). This is what I have called “sapience.”
Consciousness of self and of self being conscious is the new capacity of the human brain, evolved from precursors in mammalia, primates, and hominid genera. Humans are a new kind of animal and are potentially positioned to become a hyper-social being, not like ants or even naked mole rats, but wholly unique in how they can organize their social conditions and their lives. However, it is looking like that will take more evolution and more time than we may have available..
Observing the Observer Observing another Observer
The hierarchical structure of the brain presented here suggests that rather than a recurrent loop (Hofstadter's strange loop) giving rise to the ‘I’ in my mind, that there is something like an ultimate observer that also acts as a very long time-scale coordinator, a strategic planner that is also constructing an extended model of the self which is the basis for the sense of ‘I.’ However I think Hofstadter is also right! In my next attempt to try to understand this phenomenon of consciousness I will explore how language supports the reentrant process of the self talking to the self, narrating the autobiography of self as it develops. That autobiography necessarily includes the narrative of the environment experienced as well as an interpretation of the self and its reaction to the environment (as well as to itself). In this sense I am a strange loop that results from an evolved hierarchy of cybernetic controls.
I am also interested in the way in which we empathize with others by creating strange loops that allow us to vaguely be another person. You are a strange loop that I can kind of understand.
Damasio, Antonio (1994). Descartes' Error: Emotion, Reason, and the Human Brain, HarperCollins Publisher, New York.
Damasio, Antonio (2000). The Feeling of What Happens, Mariner Books, New York.
Fuster, Joaquín (1999). Memory in the Cerebral Cortex, The MIT Press, Cambridge MA.
Hofstadter, Douglas (2007). I Am a Strange Loop, Basic Books, New York.
. A model is a dynamic map, that is a mapping from inputs to outputs that changes with the incorporation of new knowledge and feedback from experience results.
. Actually a lowly octopus can do this trick even when there is a cork in the bottle. However, the work that the octopus does appears to be largely trial and error efforts rather than reasoned based on prior experience. Octopi have some longer-term memory retention in that if they are presented with the same problem a few hours later, their time to completion goes down. But several days later they are back at trial and error as if they had forgotten the prior lesson. And, they had. Contrast that with a crow who upon discovering a method for getting a food morsel out of a container, will retain the memory for months, if not years.