The Origin of Holism and Memory in Nature: The Systemic Memory Hypothesis



A persistent question in science is the existence of holism and emergent properties at all levels of nature -- from quantum physics and chemistry through biology and psychology to ecology and astrophysics. The simple dictum "the whole is greater than the sum of its parts, and its corollary, "the whole depends upon the interaction of its parts for its unique properties" together express the foundation of classical systems theory( 1) and modern complexity and chaos theory( 2) as they are employed in the physical, biological, behavioral, social and environmental sciences. The problem of the origin of holism in dynamical systems is resolved when circular feedback is viewed as recurrent feedback interaction. We review the logic that leads to the prediction that recurrent feedback interaction inexorably leads to the storage of information and energy and the creation of dynamical systemic memories in all dynamical systems at all levels.( 3-6) The complex non-linear interactions that naturally accumulate through the circulation of information and energy are holistic memories that reflect the evolving identity of systems as emerging and evolving wholes. The systemic memory hypothesis provides a plausible explanation for a host of seemingly anomalous phenomena in physics, chemistry, biology, medicine, and psychology(5,6), including memory in water (homeopathy), cellular memory in transplant patients, psychometry, cold fusion, out-of-body conscious experiences, and survival of consciousness after death.

Holism and Emergent Properties in Dynamical Systems

The root meaning of the word system, which derives from the Greek synhisanai ("to place together") is the concept of an integrated whole whose essential properties arise from the relationships between its parts. According to Miller(7), a system is:

a set of interacting units with relationships among them. The word `set' implies that the units have some common properties. These common properties are essential if the units are to interact or have relationships. The state of each unit is constrained by, conditioned by, or dependent on the state of the other units. The units are coupled. Moreover, there is at least one measure of the sum of its units which is larger than the sum of that measure of its units. (p. 16) [italics added]

A classic example of holism in a system is how atoms of hydrogen and oxygen, each with its own unique set of holistic properties, can combine to form H( 2)O, a molecule whose unique (holistic) properties are qualitatively different from the individual properties of hydrogen and oxygen, and only emerge when hydrogen and atom interact recurrently.

The same logic applies to the atoms of hydrogen and oxygen themselves. Electrons, protons, and neutrons all have their own unique set of properties. They can combine to form hydrogen or oxygen, atoms whose unique (holistic) properties are qualitatively different from the individual properties of the subatomic particles, and only emerge when the particular combination of the subatomic particles interact recurrently and create hydrogen or oxygen.

When the logic of interaction is carefully analyzed, especially interactions that are recurrent (i.e., circular feedback), the logic leads to the conclusion that a complex version of the history of the recurrent feedback interactions are stored within the inherent circulation of information and energy in systems.( 3-6) The emergence of a system, therefore, involves the storage of systemic memory, and as will become clear, systemic memory is memory of the whole. Since the logic of recurrent (circular feedback) interaction is systemic and generic, it can (and in fact, must) be applied to any system at any level.

According to Webster(8), one of the original definitions of circulation was "a series in which the same order is preserved, and things return to the same state." (p. 328) When the concept of circulation is reinterpreted in terms of dynamically changing, recurrent feedback interactions, things are not predicted to return precisely to the same state. Instead, the order that is predicted to be preserved is the evolution (accumulation) of the dynamical interactions among the parts. From this perspective, all dynamical systems are predicted to evolve over time.

Anticipation of the Systemic Memory Hypothesis

The hypothesis of systemic memory was anticipated in 1890 by William James.(9) He proposed: "When two elementary brain-processes have been active together or in immediate succession, one of them, on re-occurring, tends to propagate its excitement into the other." (p. 256)

When we substitute the words subsystems for brain-processes, the systemic memory hypothesis is anticipated. When two subsystems have been active together or in immediate succession, one of them, on re-occurring, tends to propagate its excitement into the other.

The hypothesis of systemic memory is implicit in writings of Warren McCulloch.(10) Not only did he propose in 1951 the idea of reverberatory memory, he pointed out:

The reverberating activity patterned after something that happened retains the form of the happening but loses track of when it happened. Thus it shows that there was some time at which such and such occurred. The `such and such' is the idea wrenched out of time. (p. 48)

He went on to say, "It is an eternal idea in a transitory memory wherein the form exists only so long as the reverberation endures. When that ceases, the form is no longer anywhere." (p. 48)

McCulloch did not extend his logic to physical and biological systems other than neurons. Had he done so, he would have discovered that reverberation circulating interaction) in systems is the rule, not the exception, and that according to modern quantum dynamics, reverberation persists, even at absolute zero temperature and in a vacuum.(11)

The concept of recurrent feedback interaction is implicit in the writings of Karl Pribram in his holonomic brain theory of perception and language.(12) His writings are replete with descriptions such as "local circuit interactions", "ensembles" with "iterations", "spatial interactions", "back propagations", and "cooperative interactions." For example, Pribram writes:

A microprocess is conceived in terms of ensembles of mutually interacting pre- and postsynaptic events distributed across limited extents of the dendritic network. The limits of reciprocal interaction vary as a function of input (sensory and central) to the network -- limits are not restricted to the dendritic tree of a single neuron. In fact, reciprocal interaction among pre and postsynaptic events often occurs at a distance from one another. (p. 16).

It is this capacity for reciprocal interactions to occur between highly interconnected, distributed, mutually interacting dendritic processes that allows for the emergence of holographic-like information to be stored throughout complex dendritic (network) systems. At the end of the Appendix(12) Yasue, Jibu, and Pribram note that information "comes to be stored in the new neural channel between the units A and B. The inference process thus has a procedure for enlarging the scope of inference." (p. 330) Though not emphasized by Pribram, the logic he used to explain dynamic memory formation in an interconnected, mutually interacting neural network system can be applied to dynamic memory formation in any highly interconnected, mutually interacting dynamical network system.

The modern concept of recurrent feedback loops in neural networks( 2) is actually a special case of systemic feedback cycles -- circulating recurrent feedback interactions -- in all systems at all levels. However, the fact that the logic used to explain recurrent feedback loops in neural networks can be equally applied to recurrent feedback networks in all dynamical systems at all levels is not widely appreciated. Moreover, the insight that circulating recurrent feedback interactions provide an explanation for holism in all dynamical systems has not been previously understood.

The systemic memory hypothesis occurred to Schwartz while he was a professor of psychology and psychiatry at Yale University in the early 1980's working on applications of systems theory to behavioral medicine and modern physics. In the process of attempting to explain to a seminar of students the basic mathematics of how feedback interactions worked (both negative and positive), he unexpectedly came to the realization that feedback interactions, especially cyclic, repeated feedback interactions (termed here recurrent feedback interactions), inherently involved the dynamic storage of information. Because he considered the systemic memory hypothesis too controversial to communicate at that time, he did not publish the hypothesis.

However, inspired by Russek's understanding and extension of the hypothesis to the storage of energy in energy medicine in 1996, the decision was reached to publish the logic and implications of the hypothesis with the hope that this would stimulate thoughtful debate and research designs to test the hypothesis( 3-6).

In this paper, we use the term memory to refer to the storage of information and energy in systems. In everyday experience, episodic (explicit) memory can be recalled consciously. However, substantial research in psychology documents that the storage of information can occur in the absence of awareness(13). Typically these implicit memories can not be recalled consciously. The use of the term memory in computer science (e.g., hard disk memory) and electrical engineering (e.g., DC battery memory) refers to the storage of information and energy. From a systems perspective, the phenomenon of explicit memory may be a special case of implicit memory.

The Example of the Two Tuning Forks

We illustrate the fundamental logic of how recurrent feedback interactions create systemic (holistic) memory using the example from classical physics of two tuning forks that come into resonance. Once the fundamental logic of this extremely (though deceptively) simple system is understood, the deep complexity of its application to complex systems will be self-evident.

If one tuning fork (A) is struck, a second tuning fork (B) some distance from A will begin to vibrate (resonate) especially if B is identical in shape, size, and substance to A. How is this phenomenon traditionally explained? The classical, non-systemic interpretation is to say that tuning fork A generates sound waves which reach tuning fork B, and if B naturally vibrates at a frequency similar to the frequency generated by A, B will begin to move in synchrony with A. A acts on B, and B reacts to A.

A systems interpretation requires that we reinterpret the relationship between tuning forks A and B as interactive, dynamically coupled, and cyclically connected. A does not simply act on B, A interacts with B. A and B are connected through the air. Since A and B are connected (coupled) when the energy from A begins to move B (cause vibration in B), B begins to generate a sound (energy), which returns (feeds back) to A. Therefore, B begins to cause vibration in A shortly after A begins to cause vibration in B. This implies that the functioning (behavior) of A (e.g., measured by the sound A generates) will be modulated in some complex ways by the functioning (behavior) of B (e.g., measured by the sound B generates).

To document these predicted interactive effects, it becomes necessary to measure the sound coming from both A and B simultaneously. Using highly directional microphones, one microphone can be focused on A (pointed away from B); the other microphone can be focused on B (pointed away from A). The sounds must be monitored simultaneously and displayed over time. Data need to be collected striking A in the absence of B and striking A in the presence of B. The systemic prediction is that we should observe that A's behavior is quantitatively and qualitatively different (e.g., increased dimensionally complexity) when struck in the presence of B as compared to when A is struck in the absence of B. Conversely, we should observe that B's behavior is quantitatively and qualitatively different (e.g., increased dimensionally complexity) when struck in the presence of A as compared to when B is struck in the absence of A.

This seemingly simple feedback interaction is not simple to model mathematically over time. Complex phase interactions (dynamic interference patterns) need to be calculated over time and distance, and as will become clear below, the interaction will naturally evolve over time due to the circulation and sharing of energy and information between A and B.

Tuning forks A and B can be shown to be interdependent and interactive, processing information generated by each of them as they interact. Since all systems are open to varying degrees at various times, it follows that all systems should interact informationally and energetically in complex and sometimes powerful ways.

The interaction between tuning forks A and B, or between any analogous objects, (e.g., photons, subatomic particles, atoms, molecules, cells, organs, organisms, groups of organisms, etc.) contains a profound implication. As A and B interact with each other, they literally create a memory of their interaction over time through the circulation of their information and energy. In fact, this memory is part of the expression of their interactions as a whole, and is a natural requirement for them to interact. The memory is the relationship. The relationship information is systemic information, and systemic information is an expression of the whole. Moreover, it follows that the information of the whole of a system is stored within each of the parts comprising the system to varying degrees.

We illustrate this conclusion by considering memory from the point of view of tuning fork A (Figure 1). At time 0, before A is struck, A is spontaneously vibrating as predicted by quantum dynamics (i.e., its atoms and subatomic particles are vibrating and moving in complex, interacting, and resonating ways). When A is struck at time 1, it vibrates with a frequency a1. The frequency a1 moves (at the speed of sound) to tuning fork B at time 2. B begins to vibrate at time 2, which is some complex (linear and non-linear) product of a1 and B's state at time 2 (we will, for the sake of simplicity, assume for the moment that a1 has not changed much as it travels to B). We will call this complex interactive product a1b2.

As B vibrates, the sound now returns to A at time 3. The sound that returns to A at time 3 is a complex interactive product of a1b2 plus whatever interference occurs with the continued sound coming from A at time 3 (a3). This gives us a1b2a3 that returns to A at time 3.

Now, let us hypothesize that A is influenced by this return sound at time 3 to some degree and starts to interact with A's vibration at time 3. Not only will A's vibration have changed spontaneously by time 4 (e.g., it might be decreasing -- the simplest case), but it will have further changed by the a1b2a3 feedback interaction returning to it. The resulting sound will be a4 modulated by alb2a3, or alb2a3a4.

In other words, in one complete cycle of A-B interaction, the sound coming from A at time 4 contains the complex history (both linear and non-linear) of the A-B interaction (their relationship) over the cycle. We see an image of a1, modulated by b2, interacting back with a3, reaching A, so that A at time 4 includes the complex history of the first A-B interaction in its next interaction. Meanwhile, from the perspective of B, a somewhat parallel set of interactions is also occurring.

Each cycle includes the previous information; hence the history (memory) continues to grow. As long as the units are connected (e.g., the two forks are interacting through the medium of the air), the memory trace circulating between them will be retained, albeit modified as time goes on. All things being equal, the memory trace will evolve in time through the continued circulation of the recurrent feedback interactions (even if the intensity of the recurrent feedback interactions decrease, which is the case for the simple two-tuning-fork example).

The logic and simple mathematics of the two-tuning-fork example are obviously and deliberately grossly oversimplified. We have chosen only four time points and have described the interaction only from the perspective of tuning fork A. However, the essence of the logic should be self-evident. Any time two (A + B) or more (A + B + n) things interact, information concerning their history accumulates.

Recurrent feedback interactions are systemic -- theoretically they occur between electrons, protons, and neutrons; between the two strands of DNA; or between the brain and the heart -- to name just a few. From a systems perspective, at whatever levels the systems are interacting (from the micro to the macro), the interactive history of the energy and information should be contained in a complex way. Hence, this mechanism suggests that memory will occur to varying degrees in all dynamical systems. It is a general model of stored systemic (relationship) information. It is the inherent capacity of a system to circulate interactive information and energy that enables a system to have a holistic history and therefore be whole.

It logically follows that the more rapidly recurrent feedback interactions occur in a given system, the more rapidly a stable holistic history should emerge. For example, atoms vibrate billions of times a second. Therefore, although it could be predicted that it should take a finite amount of time for a molecular holistic history to form when hydrogen and oxygen come together as H( 2)O, the time it actually takes may be a few nanoseconds.

According to systems theory, systems are always interconnected to various degrees in various ways. Hence, energy (and the information riding with the energy) is continually being exchanged and circulated to various degrees in various ways. As a result, the memories should continue naturally.

When tuning forks A and B resonate, they become a two-tuning-fork system. Each tuning fork functions as a subsystem in the two-tuning-fork system (or we can say each tuning fork is a system in the two-tuning-fork suprasystem -- the principle is the same). The two tuning forks become a whole.

Tuning forks A and B each contain molecules. Molecules are subsystems within each tuning fork system (or we can say each molecule is a system within a single tuning fork suprasystem -- the principle is the same). Hence, it follows that resonance not only can occur between tuning forks A and B, resonance can occur within tuning forks A and B. In fact, tuning forks A and B cannot vibrate as individual tuning forks unless their molecules can vibrate (resonate) interactive within each tuning fork as a whole. In other words, the logic of what happens between tuning forks A and B applies to what happens within tuning forks A and B as well.

Recurrent resonance, therefore, not only occurs between physical systems (e.g., between tuning forks A and B), but also occurs within physical systems (e.g., within tuning fork A and within tuning fork B) as well. For this reason, the logic that leads to the hypothesis that recurrent interaction creates memory between tuning forks A and B, also requires that we entertain the hypothesis that recurrent feedback interaction simultaneously creates memory within tuning fork A and within tuning fork B as well, and this intra-tuning fork memory is sustained, even after tuning forks A and B have been separated. The same logic requires that we entertain the hypothesis that once hydrogen and oxygen have interacted recurrently as H( 2)O, if hydrogen and oxygen are subsequently separated, some version of their history as H( 2)O will be retained within the hydrogen and oxygen, expressed potentially in terms of their individual, dimensional complexities.

Modern quantum physics indicates that even at the temperature of absolute zero, matter vibrates, and hence, resonates(11). Quantum mechanical fluctuation energy of the atoms in matter has been measured by noting the vibrations in a crystal as its temperature is lowered. The experimental data agree with the predictions of the equations of quantum mechanics, suggesting that quantum mechanical zero-temperature vibrational fluctuations of atoms in matter are a general property of matter. For example, residual quantum mechanical vibrational energy is used to explain why liquid helium does not freeze even when it is cooled to within microdegrees of absolute zero temperature. For the sake of completeness, it should be noted that modern quantum physics also suggests that a vacuum can sustain an infinity of electromagnetic vibrations (the quantum mechanical electromagnetic fluctuations of the vacuum). Hence, recurrent feedback interactions not only exist within matter, but it would be predicted to occur within the vacuum as well.(11) It logically follows that the systemic memory hypothesis can be extended to the vacuum itself with the inherent parallel creation of virtual dynamical energy systems (i.e., in the vacuum between the A's and B's shown in Figure 2) that continue the integrity of the dynamical material systems (even after a given material system may be destroyed(6)).

By definition, systems can potentially store only what they are capable of responding to (and hence processing), and they will process this information in their own ways. It follows that the nature of information stored between subatomic particles, for example, will be of a different order from that stored between neurons. Also, the more reliable and flexible the components of the system (and hence, the more complex the system), the more reliable and flexible should be the storage (and complexity) of the information.


Of course, outlining the logic that memory is intrinsically created and stored in systems does not imply that this information, once stored, can be accessed and retrieved (at least in human consciousness). Everyday experience and substantial empirical research reminds us that our ability to recognize information is typically far greater than our ability to recall information.(13) Clearly, failure to observe recall does not necessarily imply that memory has not occurred. Forgetting, therefore, does not necessarily imply that memory has been erased; the process may involve an alteration in retrieval. The deep question of retrieving memory, once stored in systems, remains a central challenge for future science. The solution may require a deep understanding of recurrent resonance as a systemic recurrent pattern recognition process.


A controversial implication of the hypothesis of systemic memory is the prediction that information, once received, is retained in some form forever, so long as the system remains intact and recurrent feedback interactions (cycling) continue. Not only will the information continue, but it potentially will evolve over time in the vacuum of space. In fact, in a deep sense, it may be virtually impossible to erase information completely in an intact system. Only in the case of presumed complete entropy Would it be predicted that systemic memory should be eliminated completely (note that entropy is a concept derived from classical physics -- systemic physics would predict that recurrent feedback interactions would likely continue in the entropic state( 4)). Theoretically, if the experiment is sensitive enough, evidence for savings or other subtle measures of change in the functioning or behavior of a system should be demonstrable in all systems as a function of the evolution of the hypothesized memory process.

Some Implications of the Systemic Memory Hypothesis

In conclusion, the wholeness of a system may derive from its capacity not only to interact, but to interact recurrently, and to circulate (and therefore mix and accumulate) this information and energy within the system. Though neurons are especially gifted in storing sensory and psychological information (because neurons are so highly interconnected, creating profoundly complex recurrent feedback interactive networks), it may be time to alter our intellectual heritage (which understandably encouraged us to adopt a kind of neural chauvinism) and re-envision the brain as being a marvelous special case of a ubiquitous systemic (holistic) memory process at all levels of nature.

It is our belief that in the process of researching the phenomenon of recurrent feedback interactions at all levels of nature, science will gain a deeper understanding of the essence of holism and evolution and in the process will enable us to understand certain heretofore unexplained observations that strain our current models of how memory works. Examples of seemingly anomalous phenomena potentially explained by the systemic memory hypothesi(5,6) include memory in water (homeopathy), purported cellular memory in transplant patients, psychometry, so-called cold fusion, morphic resonance, out-of-body conscious experiences, and survival of consciousness after death. The systemic memory hypothesis addresses many challenging questions in current frontier science, including how living systems can be viewed as dynamic macroscopic quantum systems,(14) and how "dead" matter can be "aware" of its environment.(15)

(1.) Bertalanffy, L von., 1968. General System Theory. New York, NY:Braziller.

(2.) Kauffman, S.A., 1993. The Origins of Order. New York, NY: Oxford University Press, 1993,

(3.) Schwartz, G. E., Russek, L.G., 1997. Dynamical energy systems and modern physics: Fostering the science and spirit of complementary and alternative medicine. Alternative Therapies in Health and Illness. 3,3: 46-56.

(4.) Schwartz, G. E., Russek, L. G., 1997. The challenge of one medicine: Theories of health and eight "world hypotheses." Advances: The Journal of Mind-Body Health. 13,3: 7-23.

The Origin of Holism and Memory in Nature

(5.) Schwartz, G. E., Russek, L. G., 1998. Do all dynamical systems have memory? Implications of the systemic memory hypothesis for science and society. In: Pribram KH (Ed.). Brain and Values. Hillsdale, NJ: Lawrence Erlbaum Associates.

(6.) Schwartz, G.E.R, Russek, L. G. S., 1998 in press. The plausibility of homeopathy: The systemic memory mechanism. Integrative Medicine.

(7.) Miller, J.G., 1978. Living Systems. New York, NY: McGraw-Hill, 1978.

(8.) Webster, N., 1997. Webster's New Twentieth Century Dictionary of the English Language. Unabridged, 2(nd) Edition. New York, NY: Collins World.

(9.) James, W., 1890. Psychology (Briefer Course). New York, NY: Holt.

(10.) McCulloch, W.S., 1951. Why the mind is in the head. In: Jeffress LA, (Ed.). Cerebral Mechanisms in Behavior. New York, NY: John Wiley, 1951.

(11.) Milonni, P.W., 1994. The Quantum Vacuum. New York, NY: Academic Press, 1994

(12.) Pribram, K.H., 1991. Brain and Perception. Hillsdale, NJ: Lawrence Erlbaum Associates.

(13.) Schacter, D.L., 1996. Searching for Memory. New York, NY: Basic Books.

(14.) Smith, C.W.,1998. Is a living system a macroscopic quantum system? Frontier Perspectives, 7,1: 9-15.

(15.) Graneau, P., 1998. Is dead matter aware of its environment? Frontier Perspectives. 7,1: 16-23.

The Center for Frontier Sciences.


By Gary E.R. Schwartz and Linda G.S. Russek

Share this with your friends