3.3 Information Theory View of Memes
3.3.2 The Dynamic Environment
Once transmitted, a message enters the environment through some medium or mediums. The environment interacts with messages dynamically through natural physical processes, introducing random and systemic noise. Research in linguistics (Dell, Chang, & Griffin, 1999), marketing (Costley, Das, & Brucks, 1997), and communication (Schramm, 1963) examine the implications of different message encodings and mediums on messages. All results pertinent to generic physical transmissions must apply to memes, which comprise a subset of messages.
The field of message passing cannot be done justice in a short section, so only the most fundamental processes will be examined. For a full treatment, Heath and Bryant(2000) provides an overview of transmission processes and protocols. Mediums have limited transmission rates and can superimpose dynamic noise onto a message. Message size, message fidelity, and signal to noise ratio significantly affect the variation and replication ability of memes (HaleEvans, 2006).
Message size influences the ability of a meme to propagate. Longer messages require greater transmission and reception times. The complete text of Hamlet reproduces more slowly than the phrase “To be or not to be.” Long transmission times utilize more behavioral resources. This can reduce the opportunity and motivation to repeat a message. Production of signs such as books and mass media mitigate the behavioral constraints, as a single reproduction lasts longer.
For longer memes that cannot be expressed through a lasting medium such as print or photography, transmission can be extremely difficult. The difference between choreography and notational music highlights this issue. Sheet music preserves the sounds of composers such as Mozart and Bach, which can be expected to last for hundreds of years. Choreography lacks a universal format for recording movements, forcing complex dances to be taught through direct instruction or video presentation (HutchinsonGuest,1989). This makes copying long sequences of choreographic movements time consuming, meaning that choreography is often lost once it is no longer performed (J. Anderson,1982).
Different mediums vary by the characteristics of the noise introduced. Degradation (d) and competition (c) introduce noise into a signal through different mechanisms. Assume that the meme transmission carries the signal as a functionT Rof the meme informationx. Equation3.2expresses the received message (Ra) by an agenta. All processes operate over time, expressed ast.
Ra(t) =T Rx(x, t) +d(t) +c(t) (3.2) Degradation reduces message quality due to entropy and other changes to the medium. Degradation differs for written and spoken forms of words- the waveform of a spoken word loses coherence more quickly than a printed word. The rate of degradation affects how long signs continue to present the physical information of a meme. Degradation may not represent a fully random process. Writing your name in the sand by the ocean will certainly degrade overnight, but not uniformly.
Competition introduces non-random noise that carries meaningful semantic information. Competition within a medium increases noise by superimposing an alternative signal. Competition within a medium can directly affect the physical value of a signal. Superimposing analog electrical signals creates a new signal, which may cancel out pieces of the original signal. Separate signals can be disentangled by relying on orthogonality of characteristics and error correction. FM radio employs this principle (Carlson, 1981). Simultaneous voices speaking follow this pattern.
Signals may combine such that the original signals are irretrievable. In this case, competition destroys components of one or both original signals. Changing the voicemail message on a phone deletes the old message, for instance. Entropy created by a large number of competing signals can also create this effect by introducing enough small errors as to be irrecoverable. A sufficiently large number of voices in a room can create this effect.
Noise and error correction introduce variation for memes. One criticism of memetics contends that this causes memes to vary too much to retain meaning (Atran,2001). Considering that a word in language can be treated as a meme, this criticism lacks general appeal. Research on information theory has shown that
variation can be reduced through messages with less information per length and coding methods such as parity. Discretized transmissions also have advantages over continuous transmissions; they reduce the set of allowable states in a signal. However, the introduction of variation does require careful definition of a meme for empirical study- a point addressed later in Section4.1.
Degradation and competition are two sources of noise for a meme. The signal to noise ratio has important implications for a meme. Meme transmissions will encounter noise and may undergo error correction, depending on the medium and agents involved. Error correction and handling of incompletely received messages is a significant issue which will affect some memes, but it appears unclear what theories of human cognition imply for memes. Related work in linguistics (completing words and phrases) and visual perception (picture completion) has implications for this, but these are not sufficient. The most difficult question on this matter appears to be how information is handled when it is transmitted in multiple partially observed sessions. To this researcher’s knowledge, a consensus mechanism of this mechanism does not exist. Accordingly, this conceptual model lacks a specific theory for handling incomplete meme transmission, forcing the problem to be addressed anew for any meme where this mechanism significantly influences dynamics.
Stimulation as Reception
Stimulation determines what environmental stimuli an agent physically detects through its bandwidth of detection channels. Sensory organs provide the set of stimuli for an agent and are necessary to receive a meme. While important biologically, this mechanism holds limited value for establishing the fitness of memes within a society. The sensitivity of sensory organs can provide insight into sources of noise and sensory error correction mechanisms. Significant empirical work has mapped out minimal detectable stimuli for human sensations, such as light detection and weight approximation (Levine & Shefner,1991).
Being able to physically sense the signs of a meme provides a limited form of pruning out transmissions. While the ability to sense stimuli is necessary, sensation alone is not sufficient. Stimulation provides no insight into the success of a meme with regard to its semantic information. Failing to learn the lyrics to a song because the pitch fell outside one’s hearing range provides meager insight into the workings of society.
Detection of stimuli conveying a meme is a necessary condition for meme transmission. This condition is somewhat lax, however. A variety of modes of transmission may exist and only one viable transmission mode must exist. The syntax-semantics divide allows for multiple representations of the same content (Halliday,1978). Practically, this means that a possible meme can be ruled out only when the meme cannot be transmitted using any combination of available
behavior and perception. Boundaries separating agents, such as spatial distance or language barriers, create this effect. Behavior, memes, and social factors will control some of these boundaries. For example, free individuals choose the context and persons they interact with. For the first order examination, these barriers can be taken as external to the model. This is because at any particular point in time, the physical barriers operate independently of their cause.
When agents interact, even in a limited manner, disproving the possibility of a physical transmission requires the proof of a negative: that no syntactic representation can spread the semantic information socially. Incommunicable learning such as the experience of color (qualia) and muscle memory exist, but these cases approach questions outside the useful scope of meme analysis. While it is important to establish conditions which debunk the existence of a meme, sensory constraints tell little about the vast possibilities for social learning in a human population. At the first order, physical barriers are a straightforward filter on the ability to detect a meme.