|Walter J. Freeman (2007), Scholarpedia, 2(2):1337.||doi:10.4249/scholarpedia.1337||revision #123821 [link to/cite this article]|
Intentionality is the circular process of generalization/abstraction of input and specification/concretization of output by which brains achieve understanding of their environments through the cycle of prediction, action, sensation, perception, and assimilation by learning.
Intentionality is a conceptual tool that is used by many researchers with differing aims and backgrounds to explain how human beings and animals engage the world and come to understand it well enough to meet their needs. The term originated in a biological context that has been regained in one of its current meanings. However, it has acquired several other meanings in other contexts. Here we approach an understanding of the term by describing its history, its current meanings, its neurobiology, and its importance for clarifying a fundamental philosophical issue now emerging in computational neuroscience: the cognitivist subject-object cleavage vs. the pragmatist holistic view of the action-perception cycle in descriptions of how brains work.
The history of intentionality
Intentionality was introduced by St. Thomas Aquinas (1272/1952) in the 13th century to Christianize the biological doctrine of Aristotle. He used the concept to describe the process by which human beings and animals thrust their bodies into the world ("intendere", Latin for "to stretch forth" from the Latin word for "bow string", whence "tendon"), adjust to the consequences of the action by accommodating to the sites of impact and then change themselves (brain and body) to assimilate ("adequatio", adequacy), thereby coming to know the world through the adjustments needed to conform. He conceived sensory events as unique and therefore unknowable. Likewise unknowable were the initial responses in body and brain function to stimuli that he termed "phantasms" that replaced the sensory input. In contemporary terms the patterns of sensory stimuli are replaced by patterns of action potentials, the raw sense data of psychologists. He conceived that the function of imagination (phantasia) in Thomist first intentionality was to generalize and abstract over multiple experiences with stimuli in ever-changing relationships with the body as it moved through the environment. This faculty was shared with animals. The function of imagination in his second intentionality, unique to humans, was the emergence of the self that could comprehend itself in the immanent action of understanding without overt bodily action and exercise intent with the will. His doctrine provided the foundation for bioscience, medicine, law, and all other fields of intellect in Western Europe for the next 400 years leading to the Renaissance.
In the 17th century Rene Descartes (1596-1650) re-conceived the brain as constructing abstract Platonic images and symbols of the environment and operating on them in accordance with the rules of mathematics and logic. He conceived the body as a machine working in accordance with the laws of physics. He replaced phantasms with representations and replaced intention and the will with the soul. He denied that the soul controls the body as a pilot controls a ship and settled for parallel deterministic motions similar to those of Leibnizian monads. He conceived the interface as the pineal body acting as a ball valve at the entrance to the third ventricle. In the 18th century Sir Thomas Willis (1621-1675) furthered the Cartesian revolution by replacing the soul with animal spirits and distinguishing between voluntary vs. reflex behaviors. In the 19th century Hermann von Helmholtz (1821-1894) replaced animal spirits with nerve energy after proving the validity of the First Law of Thermodynamics for biological systems. Early in the 20th century Franz Brentano (1838-1917) and Edmund Husserl (1859-1938) reintroduced intentionality but with a different meaning. Husserl and his followers retained the Cartesian concept of representation along with the distinction between subject and object, and they used the Husserlian concept of intention to distinguish between a human who could understand an object and a machine that could not. The difference was that both human and machine could represent an object with a symbol, but only the human could intend the object and thereby know it. In mid-20th century Martin Heidegger (1889-1976) reintroduced the original Thomist meaning of intentionality, and Maurice Merleau-Ponty (1908-1961) realized it in the biological contexts of experimental neurophysiology and clinical neurology.
Contemporary meanings of intentionality
The key problem for all forms of natural philosophy including neurocomputational science is to explain the neural mechanisms of the rapid transposition between material energy and abstract concept. A sensory stimulus activates a different subset in an array of receptors in the eye, ear and nose with each presentation. The knowledge accrued in the brain is of the stimulus as a class; the specific receptor subsets on successive trials are unknown and unknowable. The motor trajectory such as a signature or a tennis serve is easily performed but never twice the same way; what is invariant is the class of the movement. The distinction resides in the transposition between the specific and the generic of sensation into perception, decision into action. It is not, as is commonly supposed, between the neural and the psychic aspects, which are common to both components of the transpositions but derive from different premises and techniques of inquiry.
Researchers of three persuasions seek to solve the problem with neurocomputation. Materialists conceive the problem in terms of information processing. They seek rate and frequency codes that express information in the interspike intervals of trains of action potentials induced by stimuli or driving muscles and carried by topographically organized axons. Cognitivists conceive the problem in terms of symbol grounding. They search for symbolic codes in firings of hierarchically organized feature-detector neurons representing phonemes, lines, odorants, pressures, etc., that object-detector neurons bind into representations of events at the level of words, eidetic images, and elemental actions. Pragmatists conceive it in terms of engagement between brain, body and environment. They look for neural correlates of stimuli and associated behaviors in spatial patterns of oscillatory fields of dendritic activity that self-organize and evolve as trajectories through high-dimensional brain state space. The codes are in landscapes of chaotic attractors. Unlike codes in DNA and the periodic table, these three codes have neither alphabet nor syntax. Yet experimentalists and engineers require them as epistemological metaphors in order to measure neural activity and model brain function in behavior.
Researchers in all three approaches use the term intentionality but with different meanings. Among materialists, surgeons use it to denote the biological process of healing by which a wound closes and the body re-establishes its integrity. Psychologists use it to denote purpose, commonly conflating it with motivation. Lawyers use it to denote the planning of an action and conceive the motive as the reason for an action given by or imputed to a perpetrator, who can be convicted for intent but not for motive, as in hate crimes. Functionalists and analytic philosophers use it to denote the relation between the symbol and what the symbol represents: a thought, belief or feeling is about something, hence the synonym "aboutness". Pragmatists, dynamicists and phenomenologists use it to denote the process by which a human or animal conceives some future state, plans action to attain it, predicts the sensory consequences of the action, acts, senses, and up-dates the prediction by changing the self in assimilating to the environment as it impacts the senses.
The term is often used in association with another multivalent term: consciousness. For the materialist consciousness is commonly regarded as epiphenomenal, a side effect. For cognitivists consciousness must precede Husserlian intention in order that there be mental events that can represent the relevant features of the environment. For pragmatists most intentional actions and perceptions are unconscious or pre-conscious. Only a small part of the flow of neural activity in and through brains emerges into consciousness, yet that fraction is vital for the welfare of self and society.
Considering the multiplicity of its meanings among differing researchers, one might hold that the term should be consigned to oblivion as hopelessly confusing. To the contrary, the term is an essential tool that all researchers need to distinguish between behaviors that all medical textbooks describe as either reflex or voluntary. The cognitive (and legal) distinction between the adjectives Cartesian “reflex” and Husserlian “intentional” is usually that voluntary actions are intentional and conscious, while reflex actions are not; the pragmatic distinction of “reflex” from Thomist “intention” is that voluntary actions are conscious, whereas most intentional actions are not conscious but not reflexive either.
The neurobiology of intentionality
The intentional arc in terms of neurodynamics embodies the concept of brain function held by pragmatists and phenomenologists. The alternative concept held by materialists and cognitivists is the reflex arc, which is commonly thought to begin with the delivery of a stimulus to the receptors and to be completed with the performance of the response. Measurements of the input functions, the intervening state variables and patterns, and the motor output functions then provide the numerical foundations needed to construct models of the successive transformations of neural activity patterns in terms of information processing and symbolic dynamics.
Actually, the reflex arc begins in the minds of the investigators intentionally giving the stimulus, and it closes there with their perceptions of the response. What therefore remains to be explained is the physiological origin of that outward thrust of intention by the investigators and their subjects, which Darwin called “nerve force”; Hughlings Jackson called “nerve energy”; Freud called “neuronic inertia” before fictionalizing it as “Id”; Bergson called “élan vitale”; Heidegger called “Dasein”; and psychologists have called drive, motivation, instinct, curiosity or desire, however it is shaped by the neural circuits schematized in Fig. 1. Studies in the origin of ‘spontaneous’ background activity of cortex (Freeman, 2006a and preceding) and its stability properties (Freeman, 1975/2004) source it in the mutual excitation among neurons, which mid-twentieth century theorists and neuropsychologists reluctantly called “reverberatory circuits” (Donderi, 1999; Orbach, 1999), and which, while self-stabilized within brains, can spill over into the restlessness that characterizes investigators and other explorers in great diversity.
The simplified schematic flow diagram of activity ( Figure 1) configures the main components of the intentional arc in the mammalian forebrain by which purposive behaviors are generated. The global interaction of the motor, sensory and associational areas creates a spatiotemporal pattern that is conceived to express the present brain state (Freeman, 1995). Emergence of the pattern requires participation of the allocortex, in particular the hippocampus in the space-time loop by which behavior is oriented in space by the cognitive map and in sequence by short-term memory (O'Keefe and Nadel, 1978). The present state evolves into a prediction of a future state that contains within it a plan of action to achieve that state.
The spatiotemporal pattern that implements the plan is transmitted by cortical neurons into the brain stem and spinal cord, with feedback via corticostriatal, corticocerebellar and corticothalamic feedback loops constituting the control loop. Additional feedback is delivered through the proprioceptive loop of the body that monitors action and evaluates the performance with respect to the intent. These loops serve to specify and concretize the generic intended action. Simultaneously the sensory consequences of the intended action are predicted by corollary discharges sent to all sensory cortices via the preafferent loop. These preafferent messages shape their attractor landscapes to embody the classes of expected input in the latent array of facilitated excitabilities having the form of basins of attraction.
The patterns of energy released in sensory receptors by the engagement of the body with the environment in the motor loop are replaced by patterns of action potentials, which on injection into the sensory cortical dynamics select the most relevant basin in each cortex. Convergence to the attractor in the basin enacts generalization and abstraction in the formation of a pattern, which the cortex transmits while modifying the attractor by synaptic changes with learning, assimilating in up-dating to the ever-changing environment and body. The signals conveying the selected classes of input from all sensory cortices converge to the entorhinal cortex and are unified into a Gestalt (Kohler, 1940) on passage through the space-time loop and broadcast to all areas, whereby the global state of the forebrain is updated, and the action-perception cycle is completed.
An important caveat regarding the basin-attractor-landscape concept is that it is a metaphor adapted from venerable areas of topology that may encase it with a degree of stationarity and rigidity that is inappropriate for describing neocortical dynamics. Such is the position of Tsuda (2001) and the “Gang of Five”, who introduced the more flexible term “chaotic itinerancy” among “attractor ruins”, and Bressler and Kelso (2001) who authored the concepts of “coordination dynamics” and fluid “metastability” in dynamic state space. However provisional, retention of the concept of a dynamic “attractor landscape” appears warranted for describing four properties of the perceptual process not otherwise easily encompassed. (a) A landscape exists not as a memory store but in potentia as a flexible array of possible expected outcomes of an act of observation, only one of which is realized when the corresponding basin is selected by input. (b) The emergence of the landscape is observable by a sequence of events constituting a phase transition through a discontinuity that lends itself to description as crossing a separatrix. (c) The descent into a basin of attraction readily executes the otherwise inexplicable processes of abstraction and generalization. (d) The vanishing of the entire landscape immediately following readout of an attractor handily solves the problem faced by conventional usage requiring large expenditure of energy to get the system out of the deep well of whichever attractor basin was selected (Freeman, 2006b).
The philosophy of intentionality
Heidegger (1975/1988) reintroduced what he called "the enigmatic phenomenon of intentionality" in forms close to those of Aquinas and addressed "the central problem of philosophy" with which we began: in his terms, "... the transposition [transcendence] of the Dasein over to things. ... It will turn out that intentionality is founded in the Dasein's transcendence and is possible solely for this reason --- that transcendence cannot conversely be explained in terms of intentionality" [p. 162]. He conceived the Dasein as neither objective nor subjective, and in those terms dealt with two "misinterpretations". First was the common sense assignment by cognitivists of intentionality to the subject causing perception of an object by making a representation [Searle (1983) would say by causing the firing of feature-detector neurons)}, thus maintaining the Cartesian subject-object separation that is inherent in representationalism. Heidegger wrote that this view characterized "... intentionality as an extant relation between two things extant, a psychological subject and a physical object. The nature as well as the mode of being of intentionality is completely missed. ... The intentional relation to the object does not first fall to the subject with and by means of the extantness of the object; rather, the subject is structured intentionally within itself. ... [I]ntentionality is not an objective, extant relation between two things extant but the comportmental character of comporting, a determination of the subject" [pp. 60-61]. The second misconception was that "the usual conception of intentionality misunderstands that toward which --- in the case of perception -- the perceiving directs itself. Accordingly it also misconstrues the structure of the self-directedness-toward, the intentio. This misinterpretation lies in an erroneous subjectivization of intentionality. ... Intentionality is neither objective nor subjective in the usual sense, although it is certainly both ...". [pp. 63-65]. This misconception is common among psychologists who conceive intention as a mental state of goal-directedness or purposiveness.
Here again is the core problem: understanding the relation between the generalizations in the self vs. the material engagements with the world. How are classes known through unknowable likenesses: the phantasms of Aquinas, the action potentials of neurobiologists, the raw sense data of psychologists? The dynamical view proposes that a self-similar hierarchy of patterns, emerging from the structures of knowledge that are stored in the synaptic tissues of the brain, is continually assimilated through interactions of the brain with the multiple environments of the body and world.
In some deep sense this patterned activity expresses the being that Heidegger conceived as the Dasein, but at present with a significant limitation that constrains use of Thomist intentionality to describing only processes that animals and intentional robots (Kozma and Freeman, 2003) share with children still too young to remember their lives or to distinguish themselves from other intentional beings. High-level intentionality, in which the self reflects on the process of comprehending the likenesses provided by sensory processing in low-level intentionality, is barely touched by neurodynamicists, despite major efforts to explore the neurobiology of consciousness and attention. Operationally the capability for high-level intentionality is defined by the mirror test: toddlers in front of a mirror look behind it to see who is there; a few months later they watch themselves touching themselves. Present evidence for the neurodynamics of intentionality comes only from animals that cannot pass the test. New methods for macroscopic EEG and brain image analysis may meet the challenge and enter the domain of phenomenology. Hubert Dreyfus (2006) describes remarkably close correspondences between nonlinear brain dynamics and the basic conceptions of the dynamics of intentional behaviors as conceived by Aquinas, Heidegger and Merleau-Ponty (1942/1963). However, phenomenologists can only begin with consciousness of concepts that emerge far above the raw sense data. Owing to their entry at a high level, they cannot reach down to the level of sensation so as to distinguish between sensation and perception, as neurophysiologists can and do distinguish them by virtue of recordings of microscopic activity.
- Aquinas, St. Thomas (1272/1952) "Treatise on Man" in: The Summa Theologica. Trans: Fathers of the English Dominican Province. Revised, Daniel J Sullivan. Publ. William Benton, Volume 19 in the Great Books Series. Chicago: Encyclopedia Britannica, Inc.
- Bressler S.L., Kelso JAS (2001) Cortical coordination dynamics and cognition. Trends Cogn Sci 5: 2-36.
- Donderi, Don C. (1999) The Unlearned Reverberatory Circuit: Lashley's Legacy to Hebb, and How Hebb Invested it, Psycoloquy: 10,#61 Lashley Hebb (6)
- Dreyfus H. (2006) Why Heideggerian AI failed and how fixing it would require making it more Heideggerian. Inquiry, to be published.
- Freeman W.J. (1995) Societies of Brains. A Study in the Neuroscience of Love and Hate. Mahwah NJ: Lawrence Erlbaum.
- Freeman W.J. (1975/2004) Mass Action in the Nervous System" New York: Academic. © 2004: (online)
- Freeman W.J. (2006a) Origin, structure, and role of background EEG activity. Part 4. Neural frame simulation. Clinical Neurophysiology 117/3: 572-589.
- Freeman W.J. (2006b) Definitions of state variables and state space for brain-computer interface. Part 1. Multiple hierarchical levels of brain function. Cognitive Neurodynamics 1(1): 1871-3080 (print) 1871-4099 (online)
- Heidegger M. (1975/1988) The Basic Problems of Phenomenology (rev. ed.). Hofstadter A [trans.] Bloomington IN: Indiana UP.
- Kohler W. (1940) Dynamics in Psychology. New York: Grove Press.
- Kozma R., Freeman W.J. (2003) Basic principles of the KIV model and its application to the navigation problem. J Integrat. Neurosci 2: 125-145.
- Merleau-Ponty M. (1942/1963) The Structure of Behavior. Fischer AL [trans] Boston: Beacon.
- O'Keefe J.M., Nadel L. (1978) The Hippocampus as a Cognitive Map. New York: Oxford UP.
- Orbach, Jack (1999) Hebb's Cell Assemblies; Lashley's Reverberatory Circuits , Psycoloquy: 10,#56 Lashley Hebb (5)
- Searle J.R. (1983) Intentionality. Cambridge UK: Cambridge UP.
- Tsuda I. (2001) Toward an interpretation of dynamics neural activity in terms of chaotic dynamical systems. Behav Brain Sci 24: 793-847.
- John W. Milnor (2006) Attractor. Scholarpedia, 1(11):1815.
- Edward Ott (2006) Basin of attraction. Scholarpedia, 1(8):1701.
- Valentino Braitenberg (2007) Brain. Scholarpedia, 2(11):2918.
- James Meiss (2007) Dynamical systems. Scholarpedia, 2(2):1629.
- Paul L. Nunez and [[User:|Ramesh Srinivasan]] (2007) Electroencephalogram. Scholarpedia, 2(2):1348.
- Anil K. Seth (2007) Models of consciousness. Scholarpedia, 2(1):1328.
- Florian Mormann and Christof Koch (2007) Neural correlates of consciousness. Scholarpedia, 2(12):1740.
- Philip Holmes and Eric T. Shea-Brown (2006) Stability. Scholarpedia, 1(10):1838.
- Stevan Harnad (2007) Symbol grounding problem. Scholarpedia, 2(7):2373.