Synapse weight

from Wikipedia, the free encyclopedia

Synapse weight is a central term for understanding dynamic concepts of neurophysiology , which clarify the functioning of networked neurons , similar to what connectionism tries to do . The likelihood of permanent cessation of complex nervous responsiveness of a nerve cell within a neural network or the change in its usual and stereotypical reaction to new stimuli ( afferents ) can thus be better recorded. Due to the “weight distribution” of the stimulating or inhibiting afferents at a synapse on the one hand and the so-called 'stimulus threshold' of the postsynaptic ( efferent ) neuron on the other hand, the readiness of a nerve cell to react can be described mathematically by numbers. (a)


Dynamic neurophysiological concepts have gained practical importance insofar as they are suitable for being verifiable by computer simulation . (b) Since every type of function of the CNS is determined by stimulating and inhibiting influences, the idea of ​​synapse weight as the result of self-organizing processes is of universal importance, in that not only the topical diagnosis of neurological findings is suitable for these ideas of measurable weighting of forces, but also the psychodynamics of psychological processes. The biological plausibility of the model is the subject of scientific discussion. The criticism expressed by Francis Crick (1916–2004) at the end of the 1980s can be seen as a stage in the life history of the Nobel Laureate, in which a few years later he tried to form his own comprehensive theory of mental processes. As a counter-argument to this criticism, Manfred Spitzer (* 1958) sees the general biological existence of feedback loops in nervous control loops and the successful biological training methods of error feedback that have become known since the late 1980s. (c) In computer-simulated training methods, results that were similar to those of biological experiments could also be achieved more and more frequently. Permanent attitudes in the sense of skills are u. a. desirable in learning processes and thus also in memory . However, in order to achieve learning success, it is necessary to change given neural attitudes. Network models are also of crucial importance in contemporary psychology. They make it possible to simulate psychological issues - right down to purely subjective thoughts, feelings and sensations - on the computer in order to contribute to a more precise understanding of the processes and principles involved. (d)

Classical neurophysiology

Neural network , drawn by Sigmund Freud in 1895. Incoming nerve impulses (see arrow) illustrate the dynamics by continuing in topically / spatially separated neurons and thus creating a field with a particularly energetic level , cf. a. → projection .

The phenomenon of the stimulus threshold and the summation of the smallest stimuli was already described by Gottfried Wilhelm Leibniz (1646–1716). He called these nervous processes petites perceptions . Thus the idea of ​​quantifiable nerve impulses was provided. (a) Johann Friedrich Herbart (1776–1841) advocated a concept of psychology in 1824 that is based on measurable forces. The first part of his psychological work, which continued until 1915, begins, for example, with the chapter: " On the state of ideas when they act as forces ". By suppressing an idea , " a state of mind that is completely independent of it " can be brought about . Since Herbart was convinced of the associative connection of ideas, there were practical pedagogical consequences for him. Herbart thus started from a psychological theory of learning as the foundation of his pedagogy. (b) Sigmund Freud (1856–1939) developed a metapsychological concept of economics. He used the term affect amount in this context to point out the physiological necessity of regulating the mental energy balance. Manfred Spitzer, who received Freud, (e) also pointed out the affectivity in connection with the concept of synapse weight. (f) Classical neurophysiology is also familiar with similar dynamic relationships, such as the influence of pathways , circuits (a) and feedback effects (b) on the neuronal effectors. What is new about the view of the synapse weights as a determining variable for the transmission of excitation between individual neurons is the knowledge of the systematic connection of individual neurons in a neural network, see also → neuron theory . Inhibiting and exciting components meet in a network. The resulting question is how changing settings of individual elements affect the system and its in parts constant or permanent readiness to react. The system as a whole is always in need of maintaining a certain state of equilibrium, see also → Homeostasis . Not only large nerve tracts such as the pyramidal tract are connected to various parts of the CNS, network-like neural structures, i.e. functional associations of individual nerve cells, are also known in the central nervous system, see also → Formatio reticularis , limbic system . Their importance for learning processes is considered proven.

Individual evidence

  1. a b c d e f Manfred Spitzer : Geist im Netz . Models for learning, thinking and acting. Spektrum Akademischer Verlag Heidelberg 1996, ISBN 3-8274-0109-7 :
    (a) pp. 21 ff., 29, 31 ff., 45 ff., 57, 220 on “Synapses weight”;
    (b) p. 34 ff. on head. “Computer simulation”;
    (c) p. 64 ff. on stw. “biological plausibility”;
    (d) p. 220 on stw. “psychological significance”;
    (e) p. 5 to the tax office “Sigmund Freud”;
    (f) p. 323 ff. on head. “Affectivity”.
  2. ^ Francis Crick : What mad pursuit. Basic books, New York 1988; entire treatise on Stw. "Critique of the biological plausibility of models of neural networks".
  3. ^ Francis Crick: The recent excitement about neural networks . (1989) Nature 337: 129-132; entire treatise on Stw. "Critique of the biological plausibility of models of neural networks".
  4. ^ Francis Crick: What the soul really is , Rowohlt TB, 1997, ISBN 3-499-60257-1 (English original title: The astonishing hypothesis: the scientific search for the soul . Scribner 1995).
  5. ^ D. Zipser: Modeling cortical computation with backpropagation . In: MA Gluck, DE Rumelhart (Ed.) Neuroscience and Connectionist Theory . Erlbaum, Hillsdale (NJ) 1990; Pp. 355-383.
  6. ^ D. Zipser & DE Rumelhart: The neurobiological significance of the new learning models . In: EL Schwartz (Ed.): Computational Neuroscience , pp. 192-200. MIT Press, Cambridge (MA) 1990.
  7. ^ SR Lehky, TJ Sejnowski: Network model of shape-from-shading: neural function arises from both receptive and projective fields . (1988) Nature 333: 452-454; to Stw. "computer-simulated training methods".
  8. David H. Hubel : Eye and Brain . Neurobiology of vision. Spektrum Verlag, Heidelberg 1989; to Stw. "biological training methods".
  9. a b Peter R. Hofstätter (Ed.): Psychology . The Fischer Lexicon, Fischer-Taschenbuch, Frankfurt a. M. 1972, ISBN 3-436-01159-2 :
    (a) p. 87 on Stw. “Petites perceptions” in Lemma “consciousness”;
    (b) p. 29 to Stw. "Herbart and the association-theoretical tradition" in Lemma "Association".
  10. ^ Johann Friedrich Herbart : Psychology as a science, newly founded on experience, metaphysics and mathematics . In two pieces. 1824/1825.
  11. ^ Johann Friedrich Herbart (author), G. Hartenstein (ed.): Johann Friedrich Herbart's textbook on psychology . 7th edition, Leopold Voss, Leipzig 1915.
  12. Sigmund Freud : The repression . In: Collected Works, Volume X, “Works from the years 1913-1917”, Fischer Taschenbuch, Frankfurt / M 1999, ISBN 3-596-50300-0 ; P. 255 on tax. "Affect amount".
  13. a b Robert F. Schmidt (Ed.): Outline of Neurophysiology . 3rd edition, Springer, Berlin 1979, ISBN 3-540-07827-4 :
    (a) pp. 113–115 on Stw. “Schaltkreis, hemmender”;
    (b) pp. 113–116, 217 on head. “Feedback”.