Simulation of the Negative Conclusion Bias with INFERNET

Modeling human conditional reasoning of the type “if p then q” containing negations poses a challenge for connectionism. An integrate-and-fire pulsed network (INFERNET) was used to model this type of conditional reasoning. This model also provides an insight on certain human limitations. The model is compared to empirical data, and classical explanations. Statistical analysis shows that the model performance not only surpasses classical explanations but also is the best fitting model. INFERNET simulator results are also compared to human performance. Computational simulator performance compare well with human performance and limitations. Introduction INFERNET (Sougné, 1996, 1998a, 1998b, Sougné & French, 1997) achieves variable binding through temporal synchrony of node firing. In short, when one node fires in synchrony with another, they are temporarily bound together. It has a limited Working Memory (WM) span and the content of WM is maintained by oscillations. Once a node is activated, it tends to fire rhythmically at a particular frequency. This technique is successfully able to represent n-ary predicates (Sougné, 1996), relational reasoning with multiple instantiation (Sougné, 1998a; Sougné, 1998b), working memory (Sougné & French, 1997) and conditional reasoning (Sougné, 1996). This paper shows how the model can deal with negated conditionals. Many psychological studies in the area of deductive reasoning have focused on conditional reasoning of the type “if p then q”. Of course, some logicians would deny that material implication “ ⊃” is really what human mean by “if ...then”. Nonetheless, here are transcribed rules related to material implication: modus ponens (MP) p ⊃ q, p q , and modus tollens (MT) p ⊃ q,~ q ~ p . While most humans follow modus ponens, it is different for modus tollens. People also use two inappropriate rules related to material equivalence: Denial of the antecedent (DA) p ⊃ q,~ p ~ q × , and Affirmation of the consequent (AC) p ⊃ q,q p × . Throughout this paper the “if p then q” form will be called the “major premise”, p the antecedent, q the consequent. What happens when negations are introduced into the major premise? Negation can affect the antecedent or the consequent. It produces four forms of major premises. Table 1 shows these four forms and the inferences resulting from the application of the four rules (MT, DA, AC, MT).

[1]  Jonathan St. B. T. Evans,et al.  Matching bias in conditional reasoning : Do we understand it after 25 years ? , 1998 .

[2]  L. Shastri,et al.  From simple associations to systematic reasoning: A connectionist representation of rules, variables and dynamic bindings using temporal synchrony , 1993, Behavioral and Brain Sciences.

[3]  P König,et al.  Synchronization of oscillatory neuronal responses between striate and extrastriate visual cortical areas of the cat. , 1991, Proceedings of the National Academy of Sciences of the United States of America.

[4]  W. Singer Synchronization of cortical activity and its putative role in information processing and learning. , 1993, Annual review of physiology.

[5]  O Jensen,et al.  An Oscillatory Short-Term Memory Buffer Model Can Account for Data on the Sternberg Task , 1998, The Journal of Neuroscience.

[6]  W. Singer,et al.  The Role of Neuronal Synchronization in Response Selection: A Biologically Plausible Theory of Structured Representations in the Visual Cortex , 1996, Journal of Cognitive Neuroscience.

[7]  Nick Chater,et al.  Theories of reasoning and the computational explanation of everyday inference , 1995 .

[8]  E. Kandel,et al.  Learning to modulate transmitter release: themes and variations in synaptic plasticity. , 1993, Annual review of neuroscience.

[9]  W. Singer,et al.  Neuronal assemblies: necessity, signature and detectability , 1997, Trends in Cognitive Sciences.

[10]  J E Lisman,et al.  Storage of 7 +/- 2 short-term memories in oscillatory subcycles , 1995, Science.

[11]  M. Oaksford,et al.  Reasoning with conditionals containing negated constituents. , 1992, Journal of experimental psychology. Learning, memory, and cognition.

[12]  N. Chater,et al.  RATIONAL EXPLANATION OF THE SELECTION TASK , 1996 .

[13]  N. Cowan Visual and auditory working memory capacity , 1998, Trends in Cognitive Sciences.

[14]  Christopher J. Bishop,et al.  Pulsed Neural Networks , 1998 .

[15]  W Singer,et al.  Visual feature integration and the temporal correlation hypothesis. , 1995, Annual review of neuroscience.

[16]  Robert M. French,et al.  A Neurobiologically Inspired Model of Working Memory Based on Neuronal Synchrony and Rythmicity , 1997, NCPW.

[17]  Jacques Sougné,et al.  Connectionism and the problem of multiple instantiation , 1998, Trends in Cognitive Sciences.

[18]  P. Pollard,et al.  The Influence of Logic on Conditional Reasoning Performance , 1980 .

[19]  William A. MacKay,et al.  Synchronized neuronal oscillations and their role in motor processes , 1997, Trends in Cognitive Sciences.

[20]  Roger Y. Tsien,et al.  Coincidence detection in the nervous system : molecular and cellular mechanisms : consequence for learning and memory , 1996 .

[21]  Harold J. Fletcher,et al.  Developmental Increases and Decreases in Solutions of Conditional Syllogism Problems. , 1977 .

[22]  J. S. Evans,et al.  Linguistic Factors in Reasoning , 1977 .

[23]  Jacques Sougné,et al.  Period doubling as a means of representing multiply instantiated entities , 1998 .

[24]  A. Aertsen Brain theory : spatio-temporal aspects of brain function , 1993 .

[25]  J. Rinzel,et al.  Oscillatory and bursting properties of neurons , 1998 .

[26]  Jacques Sougné,et al.  A Connectionist Model of Reflective Reasoning Using Temporal Properties of Node Firing , 1996 .

[27]  Robert M. French,et al.  The Subtlety of Sameness: A Theory and Computer Model of Analogy-Making , 1995 .

[28]  Jonathan Evans,et al.  Human Reasoning: The Psychology Of Deduction , 1993 .

[29]  Wolfgang Maass,et al.  Spiking Neurons , 1998, NC.

[30]  D. Sperber,et al.  Relevance theory explains the selection task , 1995, Cognition.