m raj u ffiBl Bn H Hffw K&itt \ASjK H inui nu HI (Oil mm LIBRARY OF THE UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN 510.84 no. 588-589 Cop c 2 Digitized by the Internet Archive in 2013 http://archive.org/details/frogdesignsimula588bose _ 5-^ UIUCDCS-R-T3-588 COO-ll+69-0231 ~y FROG: DESIGN AND SIMULATION STUDY OF A MECHANISM WHICH LEARNS SELF PRESERVATIVE REACTIONS TO ITS ENVIRONMENT by DEBASISH BOSE August, 1973 THE LIBRARY OF THE UNIVERSTY OF ILLINOIS J AIGN UIUCDCS-R-73-588 COO-ll+69-0231 FROG: DESIGN MD SIMULATION STUDY OF A MECHANISM WHICH LEARNS SELF PRESERVATIVE REACTIONS TO ITS ENVIRONMENT by DEBASISH BOSE August, 1973 DEPARTMENT OF COMPUTER SCIENCE UNIVERSITY OF ILLINOIS AT URBANA- CHAMPAIGN URBANA, ILLINOIS 6l801 This work was supported in part by Contract No. US AEC AT(ll-l)lU69 and was submitted in partial fulfillment of the requirements for the degree of Master of Science in Electrical Engineering. FROG: DESIGN AND SIMULATION STUDY OF A MECHANISM WHICH LEARNS SELF PRESERVATIVE REACTIONS TO ITS ENVIRONMENT BY DEBASISH BOSE B. Tech., Indian Institute of Technology, 1970 THESIS Submitted in partial fulfillment of the requirements for the degree of Master of Science in Electrical Engineering in the Graduate College of the University of Illinois at Urbana-Champaign , 1973 Urbana, Illinois Ill ACKNOWLEDGEMENT The author wishes to thank his advisor, Professor S. R. Ray, for suggesting this thesis topic and for his constant guidance and friendship. He is also very grateful to Professor W. J. Poppelbaum for the opportunity to work in his group and for his advice and under- standing. The author acknowledges the immense help which he derived from the work of Mr. John Kreger, who designed the first version of the system reported in this thesis. Thanks are due to Mrs. Evelyn Huxhold for the patience and skill with which she typed the thesis, and to Mark Goebel and his staff at the drafting section for the drawings. IV TABLE OF CONTENTS Page 1. INTRODUCTION ■ 1 2. THE BASIC LOGIC ELEMENTS k 2.1 The Basic Elements ____ ______ U 2.2 The Memory Elements - _____ ___ 5 3. LOGICAL DESIGN OF THE SYSTEM l8 3.1 Structure of the Memory ------------- 18 3.2 Memory Integration ----------_--___ 2k 3.3 Decision Unit 32 h. THE SIMULATOR 36 l+.l The Stimuli 36 U.2 Decay 37 k.3 The Simulator 39 5. RESULTS AND CONCLUSION 1+5 5.1 Summary of a Run of the Simulator ________ L5 5.2 Results k6 5.3 Conclusion -------------------- 53 LIST OF REFERENCES 55 APPENDICES 56 A. Trial by Trial Summary of a Run of the Simulator - 56 B. Simulator Source Deck Listing ----------- 76 C. Sample Outputs __________________ 96 LIST OF FIGURES Figure Page 2.1 T*-Element (Schematic) - 7 2.2 A Typical Chain of T*-Elements 9 2.3 Logic Element Realization of a T*-Element ------ io 3.1 Output Arrangement of a T*-Element Chain ------- 20 3.2 A Level of a Memory with Four Chains of T*-Elements - 21 3.3 Block Diagram of a Memory Plane ----------- 25 3.^ Block Diagram of the Adaptive Filter --------- 26 3.5 Logic Element Realization of a MIS ---------- 28 3.6 Block Diagram of a MIU 30 3.7 Decision Logic -------------------- 33 VI LIST OF TABLES Table Page 3.1 Criteria for Decision by a MIS ------------ 29 3.2 The Decisions 35 k.l Characteristics of the Stimuli ------------ 38 k.2 List of Procedures ------------------ ko k.3 Format of the Array RDPROB 1+2 1. INTRODUCTION FROG is an artificial intelligence scheme which behaves adap- tively with its environment. The scheme was proposed by Professor S. R. Ray. The initial design and computer simulation of FROG was done by John Kreger. The results were reported in an unpublished memorandum. The work described here is based on simulation studies, by the author, of an improved design of the mechanism. FROG, as a mechanism, is an attempt to simulate the feeding habits of a small animal. FROG is cohabited by some bugs (both good and bad tasting) and predators (encounters with whom are considered painful but not fatal). The environment of FROG is separated into a number of regions characterized by different surface properties. The machine learns to discriminate between the different stimuli by means of their visual characteristics. According to Lettvin, Maturana et al. in "What the Frog's Eye Tells the Frog's Brain," the retina of a frog contains four groups of fibers which break down every image into four main characteristics which are then transmitted to the brain. The characteristics are: 1. Sustained contrast: The presence of a sharp boundary, moving or still, with much or little contrast. 2. Net convexity: The presence of a curved boundary, darker than the background and moving on it, showing most activity if the object moves intermittently with respect to a back- ground. 3. Moving edge: Tells whether or not there is a moving bound- ary in a larger area within the field. 2 h. Net dimming: Tells how much dimming occurs in the largest area and so related to the size of the object. In our system each stimulus is assigned a value, in the range to 1, for each of the above four characteristic* These are the char- acteristics (denoted as visual.. , visual^, visual, and visual, respective- ly) from which FROG infers the good or bad taste or the pain it receives from a predator. Each bug has a value n(0 £ n £ l) for its good taste and its bad taste is put as 1 - n. A good tasting bug is one whose good taste is greater than its bad taste. A bad tasting bug is one whose bad taste is greater than its good taste. A bug is assumed to inflict no pain. By contrast the predators it encounters inflict pain to a varying degree. It is to be noted that FROG utilizes the concept of "continuous logic." According to P. N. Marinos "In deductive logic, it is invariably assumed that every proposition is either true or false... In real life, however, there are concrete situations which do not fit readily into such 2 * In a later publication by Maturana et al. the number of classes of ganglion cells was revised to five. Four of them were found to act on the visual image to perform complex analytical operations that remain invariant under changes of general illumation; the fifth class measured the light intensity. The operations of ganglion cells are briefly de- scribed by their names: Class 1. Sustained edge detection — with non-erasable holding. Class 2. Convex edge detection — with erasable holding. Class 3. Changing contrast detection. Class h. Dimming detection. Class 5- Darkness detection. In spite of this finding we have decided to work with four visual characteristics in our simulation of FROG. Our effort is to devise a cognitive system which recognizes unique combinations of contexts. By not extending the dimensions of the stimulus (or context) space to five we will only restrict the complexity of the problem without affecting the difficulty of the problem. The approach we take is perfectly general and can be extended to any dimensions. a simple scheme and the conventional dichotomy of true-false is highly- inappropriate. In other words, it is quite possible that certain situa- tions might have values other than falsehood and truth." Thus, while the only values a situation might have in deductive logic are and 1, con- tinuous logic allows a given situation to have the values 0, 1, or some number between them. This is particularly suitable for our system. For each stimulus presented, FROG receives external data of two types, visual and tactile. The four visual characteristics can assume values in the range to 1. The tactile data simply specifies the region in which the stimulus appears. When FROG attempts to feed on a stimulus, it is assumed that there are external agents to evaluate the taste or the pain inflicted and make it available to FROG. Internally FROG gen- erates hunger, fatigue and alarm signals. FROG acts like an adaptive filter. The environment sensory data, interpreted through the adaptive filter, and the somatic signals, which measure the requirements of FROG for feed and rest, are combined to decide upon a course of action. The modes of activity are (l) feed; (2) rest; (3) flee. 2. THE BASIC LOGIC ELEMENTS 2.1 The Basic Elements The basic logic elements which constitute the adaptive filter are defined below. All inputs, x. , and outputs, y, have values in the continuum [0,l] unless contraindicated. Nearest Element Name Complement (C) Minimum (m) Symbol x -*©-+■ y 1* x k ©+y Analog NOT y = 1-x Definition AND y = min (x. ) i=l X Maximum (M) V ®+y OR y = max (x. ) i=l X Product (P) Equivalence (E) Selective *Memory ( MEM ) : 1 V X 2* 1 )Sh: X 2* Sense Indicator x. y = (x )'(x ) y = x x if |x 1 -x 2 | < 0.05 f (g)-»-y Equiv. (fuzzy) y = otherwise x -+MEM)- y = x 1 , if x 2 > 0.5 y = otherwise y = 1 if x > y = otherwise y = when first powered "on." If x-j_ is the first occurrence of x : > 0, y = x| Nearest Element Name Symbol Analog Definition Current *Transmittance (T) x -,"*"©/ \"* y — Time Stored Value Output (y) initially (t=0) t = t last non- min zero input / iv u, n ;. (Stored valued) i = no. of inputs till t 1' * See the following section for further explanation. 2.2 The Memory Elements 1. MEM-element The MEM-element is initially "unset" (no value stored) and its output is zero. It "sets" to (stores) and begins to output its first nonzero input. The output of a "set" MEM-element is invariant and con- stant: At any point in time the output will "be the stored value. Sub- sequent inputs will neither change the stored value nor affect the output. 2 . T-element The T-element is also initially "unset" (i.e. no number stored) and its output zero. The T-element sets to the first nonzero input and begins to output the product — • (stored value) in a constant manner. The "stepping parameter" n is a nonzero positive integer. The next time the T-element encounters a nonzero input it sets to the new value of the input and a stepping process occurs by which the T-element 's output is 2 1 stepped to (— ) • (stored value). This process of stepping by adding — to the fraction multiplier — occurs each time a nonzero value is input to the T-element until the term — = 1. At this point the T-element outputs the full value of its currently stored contents and can be considered to have reached "full potential." Any subsequent inputs to this element will not cause the stepping process to occur but the stored value will set to every new input. At any point in time after a T-element has reached full potential its output will be its stored value. The particular value of its stepping parameter, n, is an inher- ent invariant characteristic of a T-element. In the case of the simula- tion FROG it is useful to have four variations of T-elements, having step- ping parameters of 8, 6, k and 2. This result in FROG having T-elements that come up to full potential very quickly (n = 2), very slowly (n = 8), or at moderate speed (n = h or n = 6). 3. T*-element The principal macroelement of FROG's memory is a T*-element. A T*-element , shown schematically in Figure 2.1, is constructed out of the nine continuous logic elements described in section 2.1 and some delay elements. A T*-element has a maximum of four inputs (A , A_ , A_ to A> ) for the visual characteristics of the stimuli and an input B for a non- visual characteristic. A T*-element stores a set of values of a specific combination of the four visual characteristics taken one, two, three or four at a time. It also stores all the different values of the nonvisual characteristic of one kind (e.g. all the different bad taste values) which may be associated with the particular values of the combination of visual characteristics stored. The T*-element forms a product of the minimum of the visual input values stored with the maximum of all the different nonvisual inputs stored. The output of the element depends on VISSTIM (VISUAL INPUTS N = l,2,3 0R4) NONVISUAL (NON VISUAL INPUTS) NBHOOD MODNBHOOD (MODIFIED NBHOOD) Figure 2.1 A T*-Element (Schematic) 8 how many occurrences of the given combination of visual inputs FROG has come accross and will he described later. Another input to a T*-element is a write regulating signal C. The element has two outputs: Output D is the main output of the T*- element and output E is a modification of input C. The T*-elements are connected in a row. Figure 2.2 shows three of them connected together. Each of these elements retain a specific com- bination of values of two visual inputs. The visual inputs and the input B are the same for all the elements. However, input C, the regulating input NBHOOD, is set at 1.0 for the T*-element #1 (the leftmost T*- element) while input C for T*-element #2 is output E (the modified NBHOOD ] of #1. The pattern is repeated. To understand the operation of a T*-element consider a T*- element which is the leftmost one in the 'chain' described above. Assume FROG has just "eaten" a bad tasting bug. The bug has specific values for its visual characteristics and bad taste. Assume FROG wants to record the specific combination of the values of the second and the third visual inputs (visual- and visual-). The value of the second visual input appear on the A line and that of the third visual input appear on the A line. The value of the bad taste is placed on the B input line of the chain. Now, refer to Figure 2.3 for the following discussion. Consider what happens in block 1. The S-element on the B input gates in the bad taste only if it is greater than or equal to .5* This value is input to the m-element of block 1. Another input to this m-element is the signal NBHOOD, which has been set initially at 1.0. The third input to this m-element (the complement of the output of the now "unset" MEM-element in block 3) is also 1.0. The output of this m- VISUAL 2 VISUAL 3 NONVISUAL -i >— -•- — i ►— 1 ►— J f ' f \ f \ / N£_ \f \ f V ^f— \/ \f V A a A 2 B C A X A 2 B C A X A 2 B C D E D E D E i > f < f > t Figure 2.2 A Typical Chain of T*-Elements 10 -p a I H W I * En o a o ■H I* N ■H H cd H W o •H O CM •H 11 element (which is greater than .5, since it is equal to the input bad taste) becomes an input to the M-element of block 1; the other input (the output of the still "unset" MEM-element of block 3) is 0.0. The output, then, of the M-element is the same as the input bad taste value. This output, which is >_ .5, gates in the visual inputs through the S- elements in block 2 and the taste input through the S-elements in block 3. These stimuli then "set" the MEM-elements in their respective blocks. The MEM-elements, after being "set," begin to output the values of visual characteristics and taste that are still being input to them. These values reach the E-elements, and since each of the elements has two equal inputs, they now output the values of visual characteristics (in block 2) and the taste value (in block 3) that were input to the T*-element. The m-element in block 2 now select the minimum value of the visual inputs. This value is nonzero, so the lower Si-element output in block h will be 1.0, which is input to a m-element. Since the output of the E-element in block 3 is nonzero, the upper Si-element in block h will output 1.0. So the following C-element will output 0.0 which is the other input to the m-element in the same block. The m-element will now output 0.0. The taste input which is gated in at the S-element in block 3 is delayed till the output of the m-element is reliable. The delayed taste input is now applied as x input to the S-element in block 5. Since the x input is 0.0 the taste input is not gated into the MEM-element of block 5- The block 5 is similar in structure to block 3 and its function will be apparent later. The MEM-element in block 5 is still unset. The inputs to the E-element of block 5 are both 0.0 so it outputs 0.0. The output of the E-element in block 3 is the taste value stored in block 3 and that of the 12 E-element in block 5 is 0.0. So the output of the lower M-element in block 6 is >_ .5. The output of the upper element is the maximum of the stored values in the MEM-elements in block 3 and 5; right now that would be the block 3 output since the other is unset. This value is now gated into block 7 through the S-element by the output of the lower M-element which is >_ 0. 5- One input to the block 7 is the minimum of all the visual input values stored in the MEM values of block 2. The other input is the maxi- mum of the stored values in the MEM-elements of blocks 3 and 5. Each of these inputs go to a Si-element. Since the inputs are nonzero the SI- elements begin to output 1.0. The inputs also go to a P-element which forms their product. The product is the x input to the upper S-element in block 7» The x input to this S-element is the minimum of the outputs of the above two Si-elements. Since these outputs are each 1.0, the x input to the S-element is 1.0 and the output of the P-element is gated through to the T-element. At this point, the E output of the T*-element , the modified version of the NBH00D signal becomes defined. The E output of the T*- element is 1.0 only if in block 8 the delayed C input to the element is 1.0 and there is no output from the upper S-element of block J. The 1.0 on the C input (NBH00D signal) is delayed for long enough span, y, to allow a signal to be output from the upper S-element of block 7 (i.e. to allow it to become reliable). Until this happens, since the upper input to the m-element in block 8 is zero, the E output is zero. Now, when the upper S-element of block 7 has an output value > 0, the side input to the m-element in block 8 becomes 0.0, so that even after the delayed NBH00D signal reaches the top input of the m-element, the E output continues to remain zero. 13 To return to block 7> the output from the S-element has caused the T-element to "set," i.e. store this value and begin out putting — x this value. The T-element output becomes the x input to the lower S- element in block 7. The x p input to this S-element is the minimum func- tion of (l) the SI function of the visual input to block 7 which is 1.0 and (2) the complement of the SI function of the nonvisual input to block 7 which is 0.0. Thus the x input to the lower S-element in block 7 is 0.0 with the result that the D output of the T*-element is 0.0, which is what it should be, for this is a "write" into the memory cell, not a This is a complete "write" cycle for T*-element #1, but now let us consider what happens to the still "unset" T*-element #2. Its A , A and B inputs are the same as that for the first cell, but the C input, NBH00D (before and after modification by T*-element #1 ) is now 0.0. The effect of this will be to cause zero to be output by the m-element in block 1 and since the MEM-element in block 3 is unset, the output of the M-element will be 0.0. This zero is the x input for each of the S- elements in blocks 1 and 2 with the result that the visual and nonvisual inputs to the T*-element will not be gated into the T*-element. However, let us look at the output E of this second T*-element , for it will be input for the T*-element #3. The complement of the SI function of the upper S-element in block 7 will be 1.0. This value will be one input to the m-element in block 8; the other will be the value coming out of the delay unit. However, this value, both before and after delay time, u, has elasped, will be 0.0. Therefore, after one memory cell in the chain has been written into, no "unset" T*-element to the right of that memory cell may be written into. Il4 A T*-element can be written into a second time if the visual inputs match the values stored in the MEM-elements of block 2. The taste value need not necessarily match the value stored in block 3. A "set" T*-element will always gate through any visual input and any B in- put greater than .5 to blocks 2 and 3. This will be because the output of the MEM-element in block 3 will cause M-element in block 1 to output some value >_ 0.5» which, in turn, will cause the S-element gates to be constantly open in blocks 2 and 3. Since the visual inputs match the stored values in block 1, the E-elements will output nonzero values, and so the visual input to block 7 will be nonzero. On the other side, the B input will be compared with the stored value in MEM-element of block 3. If they match, the E-element will output a nonzero value. The rest of the action will be exactly similar to what happened when the element was set for the first time. A nonzero product will be transmitted to the T-element which will increment the stepping parameter and set it to the new product which will essentially have the same value as the old product stored. If the taste input does not match the value stored then the E-element output of block 3 will be zero. This causes the first input to the m-element of block h to be 1.0. The second input is also 1.0 since the visual inputs matched. The m-element output will thus be 1.0 and will gate in the delayed taste input to block 5 which will set the MEM- element in that block to the new B input value. Block 6 is arranged to output the maximum of the two stored values. The two inputs to block 7 will now be defined and the T-element will be set to the new product (which may not necessarily be the same as the old one) and increment the stepping parameter. 15 The purpose of "block 5 is now apparent. It is to store a second value of the nonvisual input (bad taste in this case) which may be associated with a given combination of visual inputs. Block pairs like h and 5 may be repeated to provide for more different values of the non- visual input. In our simulation we did not need to provide for more than two. The T*-element will be written a third time, a fourth time and so on at the subsequent appearances of the same combination of visual inputs and a bad taste value (>_ 0.5) which may or may not match the stored values. Note that the bad taste value transmitted by block 6 to P-element in block 7 is always the maximum of all the different bad taste values encountered till then for the given visual inputs. Also note that the second input to the m-element in block h is always 0.0 except when all the visual inputs match the stored values in block 2. This prevents the input taste value (the B input) to be gated into block 5 unless the visual inputs match. Now consider what will happen if a different set of A , A input values are input to the chain. They will reach the T*-element #1 , the one that is "set." As has been said previously, a "set" T*-element always gates the A and B inputs to block 2 and 3. Since the MEM-elements are already set, the new inputs will not affect them, but will be com- pared with the MEM-element outputs at the E-elements. Since we are assuming that at least one visual input is not within allowed tolerances for the E-elements, one or more E-elements in block 2 will output zero. The visual output from block 2 will then be zero which inhibits the taste signal to go to block 5- Block 3 could output some nonzero value if the B input matched the value stored in block 3- So the block 6 output may 16 or may not "be zero. The block 2 output being zero both the S-elements in block T will be inhibited. So the D output of the leftmost T*-element will be zero. Also the E-output of this T*-element will be 1.0 after time y. So input C of the T*-element #2 is 1.0 and this will become "set" in exactly the same manner as the leftmost T*-element. The NBH00D signal functions solely to tell the chain that a T*- element has been written into for a given combination of visual inputs and, hence, once this occurs, not to write into it for a different set of visual inputs. Note that for a given combination of visual inputs the T*-elements can hold more than one B-input values. In the present simula- tion provisions are there for two such values. The NBH00D signal makes sure that the writing of a new combination proceeds sequentially down the chain from left to right. All of the above discussion pertains only to "writing" into the T*-element memory cells. Now the question arises how is a T*- element caused to output the current output state of its T-element? The process occurs contextually, when FROG has just seen a stimulus and its memory cells must be consulted to see if FROG has had any experiences with this stimulus in its past and if so, whether they were good or bad experiences. Obviously, then, no nonvisual input is presented to the memory in this case, for feeding has not yet taken place; thus, the non- visual input to a T*-element during a read cycle is zero. The "reading" process occurs in the following manner. The visual stimulus is presented to all T*-element in the chain at once (reading, it will be seen, does not have to occur sequentially, as does writing). The visual inputs will be gated into block 2 of each of the 'set' T*-element in the chain. If the current visual input values are IT close enough (to within +_ 0.05) to that which are stored in the block 2 MEM-elements, then the E-elements in block 2 will output nonzero numbers; if not, no read will occur. Wow recall that the x_ input to the lower S-element in block 7 is the minimum of two inputs (l) the SI function of the visual output from block 2, which is 1.0 and (2) the complement of the nonvisual input from block 6, which (since the B input to the T*-element is zero) is 1.0, with the result that this second input to the m-element is also 1.0. Thus, since the m-element is outputting 1.0, the current output status of the T-element is caused to be output; no stepping of the T-element will occur for there is no input to it in the read cycle. A write into an "unset" element cannot possibly occur by accident during a read cycle, for if a T*-element is "unset" and the nonvisual input is zero, then the visual stimulus input will not get past the S-elements in block 2. Also write into a block 5 of a 'set' T*-element is not possible as the B input is zero. 18 3. LOGICAL DESIGN OF THE SYSTEM y As has been mentioned earlier FROG, as a cognitive system, is like an adaptive filter, such that when the input sensory data and the somatic signals are presented a course of action for FROG is de- cided. It consists of a 'memory' where the experiences with different stimuli are stored, i.e. which develops a 'model' of the world from which the stimuli are drawn. The memory is followed by a 'memory in- tegration unit' (MIU), which is to handle the various memory outputs properly on presentation of a stimulus and to make them available to a 'decision unit' (DU), which decides on the mode of activity. 3.1 Structure of the Memory Structurally the memory of FROG is divided into several parts (memory planes) equal in number to the number of regions in the environ- ment. Experiences in the various regions are stored in separate memory planes. The memory plane for each region is again divided into three parts. The first is a good taste memory (GTM) which associates the distinguishing visual characteristics of good tasting bugs with their tastes. The second is the bad taste memory (BTM) which associates the visual characteristics of bad tasting bugs with their tastes. The third is a predator memory (PM) which associates the visual characteris- tics of a predators with the amount of pain they inflict. The association of a nonvisual input (good taste, GT, bad taste, BT, or pain inflicted, PAIN) with visual characteristics or its In the sense of psychologists; meaning, a response judgment requiring an extended-time "consideration" of the stimulus data as opposed to a recognition task in which the response is relatively rapid and automatic 19 combinations is done in a T*-element as described in Chapter 2. The question now arises as to how these T*-elements be arranged within a memory to simulate an efficient learning scheme. The strategy adapted in FROG is described below. First of all, we see if any of the four visual characteristics might be a basis for discrimination between various stimuli. For this, we take three sets of four chains of T*-elements and place a set in each memory. These T*-element have one visual input each. The D out- puts of all the T*-elements in each chain are connected to one M-element (see Figure 3-l). Thus the T*-elements in chain 1 in BTM would associ- ate first visual characteristic of a particular stimulus with its bad taste. The second, the third and the fourth chains in each memory would perform similar functions with the second, the third and the fourth visual characteristic respectively. The outputs of each of these chains in a particular memory are then input to one M-element as shown in Figure 3.2. For the purpose of simulation study a chain is made up of three T*-elements. So 3 stimuli of each type can be remembered. Let us examine the process of read out from and writing in to the memory. If a bug has just been eaten or a predator encountered, the four visual characteristics would be sent to each memory; visual charac- teristic 1 to chain 1, etc. The nonvisual inputs good taste (GT), bad taste (BT) and pain (PAIN) are also sent to the GTM, the BTM and the PM respectively. For a bad tasting bug, writing will take place in BTM only (since BT > 0.5, GT < 0.5, and PAIN = 0.0). Whereas, for a predator, writing takes place in PM only if PAIN >, 0.5 ( GT = BT = and PAIN >_ 0.5). When FROG has just seen a stimulus, a read is performed in each memory by sending the visual characteristics to their respective chains 20 VISUAL INPUT NONVISUAL INPUT NBHOOD (=1) ± ± ± ± ± ± ± ± ± M CHAIN OUTPUT Figure 3.1 Output Arrangement of a T*-Element Chain 3 Q- I- 3 o _l UJ > UJ 21 X o "* — * — *r CO < X o AAA ro < x o AAA < x u * /F - ^ -P a aj 0) H M I * Cm O •H .a, o M o En X! -P •H >> U O o OJ 00 •H 3 0- -J < 3 > 2 5 3 en > z o Q o o I 00 z C\J 3 CL 2 < 3 (/■) 3 a. z o CD 3 a. < 3 3 _ 9: - < 3 en a o O x CD 22 along with zero nonvisual inputs to all memories. Any T*-element in which the visual input is numerically within a neighborhood of 0.05 of the value of visual characteristic stored in the MEM-element in block 2 will output the status of its T-element. Then the maximum function of all the T*-outputs in the memory will be output from each memory. That is, GTM will output the numerically greatest impulse possible for eating the bug, the BTM the numerically greatest possible impulse for avoidance of the bug, and PM the numerically greatest possible alarm. These out- puts, one from each memory, are compared in a manner discussed later. A problem arises when, for example, a good tasting bug and a predator have approximately equal values for their numerically largest visual characteristic and the good taste value of one is approximately equal to the pain inflicted by the other. If any of these two stimuli is presented the GTM and the PM outputs will be approximately equal and FROG will not be able to make a choice based on its memory outputs. This problem arises because the four visual characteristics are too general to form a basis for discrimination. The natural solu- tion is to take combinations of these four basic characteristics and use them as inputs to chains. The four visual characteristics taken two at a time forms the following six combinations : (1) visual.. , visualp (2) visual , visual- (3) visual , visual, (k) visual-, visual- (5) visual-, visual, (6) visual , visual. Now we form six more chains of T*-elements in each of GTM, BTM and PM. 23 Each chain has two visual inputs and one nonvisual input, depending on the memory involved. Should the above combinations prove insufficient as a basis for discrimination between certain stimuli, combinations of the visual characteristics taken three at a time are utilized. These combinations are (l) visual., , visual^, visual,,, _ 1 , ,_„ 2> 3! (2) visual , visual-, visual,, (3) visual , visual , visual, , (h) visual , visual , visual, . Each of these combinations is input to a different chain of T*-elements in each memory, accompanied by a suitable nonvisual input. The last level of discrimination is that using the four visual characteristics taken all at a time. Thus each of the memories (GTM, BTM, PM) has 15 chains of T*- elements which may be divided into four groups referred to as "levels." Level 1 consists of four chains of T*-elements that have as their visual inputs the four basic visual characteristics of the stimulus. Level 2 consists of the six chains of T*-elements that have as their inputs the combinations of the four basic characteristics taken two at a time. Level 3 consists of the four chains of T*-elements that have as their visual inputs the basic characteristics taken three at a time. Level h consists of the chain which has as its visual inputs the combination of the four basic visual characteristics taken four at a time. Now we can justify why it is necessary to have upto four A inputs to a T*-element. Level 1 chains need T*-elements with one visual input (A ) . Level 2 chains need T*-element with two visual inputs and so on. The GTM, the 2U BTM and the PM all have the same structure with the only physical dif- ference being the nonvisual input to the memory. The level output of any level is the maximum function of all the T*-element outputs of that level. The different kinds of T*-elements not only differ in the num- ber of visual inputs that it, can handle but also in the value of n. It is at this point that this variation becomes meaningful. It is apparent that, in going from level k to level 1, there is an increasing generality in the visual inputs to the chain, level 1 being the most general and least specific and level k being the least general and most specific. Thus, the most general level should learn the slowest in order to avoid making mistakes in discrimination of a very broad nature; but as the amount of generality decreases, faster learning can be permitted be- cause a mistake in the lower levels would not be as disasterous as it would be in level 1. In level 1, the T*-element contain T-elements which have n = 8. Thus it is necessary for the visual input to repeat (within the allowed neighborhood of +_ 0.05) seven times after the initial setting of the T*- element in order to have the T-element come up to full output potential (full learning). The T*-elements in level 2 contain T-elements for which n = 6, so five repetitions are necessary for full learning. In level 3 T*-elements n = U, so three repetitions are necessary for full learning. In level k, the most specific level, the T*-elements contain T-elements for which n = 2; thus only 1 repetition of the stored visual inputs is necessary for full learning. 3.2 Memory Integration When FROG has just seen a stimulus the visual characteristics are sent to all the memories without any taste or pain (i.e. nonvisual) 25 GT BT PAIN GT GTM LEVEL I (NON -VISUAL INPUTS) BT BTM (1,1) (1,2) (FIFTEEN VISUAL INPUTS) PAIN (4,1) PM GTMOUT 1 GTMOUT 2 GTMOUT 3 GTMOUT 4 BTMOUT 1 BTMOUT 2 BTMOUT 3 BTMOUT 4 PMOUT 1 MEMORY miu PMOUT 2 PMOUT 3 PMOUT 4 ALARM OUT MIU OUT NOOEC* Figure 3.3 Block Diagram of a Memory Plane UJ UJ 26 Q UJ UJ u. 1 1 1 B UJ a TT UJ O 2 < TTTTO Rl IO ♦ sindNi nvnsiA X 3 o 2 en < M 2 > O 2 ui 2 M 2 > o 2 ui 2 CVJ (*)tO3Q0N {£)tO3Q0N v tr o 2 ui 2 (2)tO3Q0N (I)tO3a0N SN0I1VNI9W03 indNI IVDSIA ST '(NlVd '18 '19) SindNI IvnSIANON £ CU •P H •H Ph 0) > •H ■P < J=! p O •H Q O o H pq 0O (U •H 27 information. Let the stimulus be in region 1. There is a signal REGION which specifies to FROG the region in which stimulus has been presented. The read out should occur in the memory plane 1 which stores information regarding region 1. The GTM, the BTM and the PM in plane 1 now outputs twelve values; one per level, with four levels in each of the three memories. These outputs are compared and processed in a manner described below by a 'memory integration unit' (MIU) and made available to the 'decision unit' (DU). There is a MIU in each plane. The MIU in the plane compares three signals at a time, starting with the three level 1 outputs. This comparison is done in an element called the 'memory integration subunit' (MIS). The outputs from a memory plane are ALARM0UT, MIU0UT and N0DECU. (See Figures 3.3 and 3.U). If FROG senses a predator it will have an ALARMOUT. If FROG sees a bug there will be an output, MIUOUT, proportional to its desire to eat. But if no decision is reached the N0DECU output of the plane is 1.0. When this happens FROG gates in the visual inputs to all the other planes for a readout from them. This enables FROG to come to a decision based on its knowledge of this stimulus in all the regions. The maximum of the MIUOUT of all the planes is the final MIU output (MIUOUTF). The ALARMOUT outputs of all the planes are also input to a M-element, the output of which is the final alarm output (ALARMF). If no decision is reached in all the planes NODECU from all the planes will be 1.0 so NODECF, the final indication of whether a decision has been made or not, would be 1.0. The logic element realization of the MIS is shown in Figure 3.5- There are three inputs to the MIS; the three level outputs from the corresponding levels of GTM, BTM and PM in the plane (see Figure 3-3 28 (EHD d9lS L 1 1 A 1 4 1 \~ o > w _J »- OD > 2H Z> o > _J t- o l UJ _l Q. L> CO H O a o •H -P cd 03 « -p a> S _ 0.5 GTM0UT BTMOUT is the largest and >_ 0.5 BTMOUT PMOUT is the largest and _> 0.5 PMOUT None of the input is largest by definition and/or all of them are < 0.5 1 Table 3.1 Criteria for Decision by a MIS Thus it is necessary for the numerically largest input to be greater than 0.5 and greater than the next input value by at least 0.05, if a decision is to be amde. If these conditions are not met, then the outputs MISOUT and ALARMIS are zero and the output NODEC is one. A MIU consists of four MIS modules with some additional logic to connect the modules (see Figure 3.6). The first MIS compares the level 1 outputs of the three memories, if no decision is reached, based on the above criteria, then the outputs MIS0UT1 and ALARMOUTl are zero and N0DEC1 is one. This allows the level 2 outputs to be gated into a second MIS. If comparison in this MIS yields no decision the outputs MIS0UT2 and ALARM0UT2 are zero and N0DEC2 is one. The process continues 30 GTMEMOUT 1 BTMEMOUT 1 PMEMOUT 1 MIS IP u GTMEMOUT 2 BTMEMOUT 2 PMEMOUT 2 G €> ALARMIS1 MISOUT 1 NODEC 1 MIS ALARMIS2 MISOUT2 I! \> GTMEMOUT 3 BTMEMOUT 3 PMEMOUT 3 <•> GTMEMOUT 4 BTMEMOUT 4 PMEMOUT 4 0.5 or zero when hunger is < 0.5- 33 UJ 3 o ►- < Q Ul in UJ u. 6 Al a — < ► UJ Ul u. t- — 4 ► Ul J jr 1 1 jj 2 2 ^*V C -v~"\. y o I i < QJ o -© s* ^V ^^^V ^^*N. ^^^ UJ z p 9^>HgK- €>- n* o C3 1 r * X 3 Ul f t I Q UJ (E 1 Ul Li : u. J UJ Ll. A 4 » C > Q u. > o 2 : : Z or < _i < Z o h- < o o -© o o UJ QC r?^£) -© bO O c o •H W ■H CJ QJ Q tn •H 3U The output of the M-element is one of the input to the top m-element. The other inputs to it are HUNGER and the innate visual stimulus (INNATESTIM) , which is simply the maximum function of the four visual characteristics. It serves in the early trials to tell FROG whether or not what it sees is a bug before its memories have learned enough about the bugs. The output of this m-element goes to a M-element. The other input to this M-element is nonzero only when hunger reaches the point of starvation (hunger = 1.0); when FROG is starving this input assumes a value equal to the innate visual stimulus. Thus, when FROG is at the point of starvation, its feeding impulse would be fairly large, i.e. FROG will eat any bug regardless of what its memories say about its good or bad taste. When FROG is not starving, the output of the m-element determines the value of FEED. FROG can assume one of five decisions as detailed in Table 3.2. The tabulation is fairly self-explanatory. 35 CO O ■H -P CU Oj si a> -P CO P Tj 0) • 1— 1 o3 CU T3 CU H fl (0 tH p a3 a •H •h a • O . O ,G T3 CU -p 0) -H K CO K -H bO bO C h E 3 Ph O O C a3 CU ft 0) 1 o cu •H cu o a TJ « T3 -P g a3 d o S a ph CO o o •H •H •H 03 -p CU o o G co bO P •H !h -p CD cu CO CO CD -P Ph 7* a o cu p> CO 2 cu H O co h co cti Q cu •H 11) O w X! W CU cu P M -P co a3 CO P •H O 3 Cm o T) •H O o pa a •H w o w W i— i pa H -p Ph J Ph J Ph LTN Ph LfA EH Ph u^ EH •H • • < • < Td A ~tk A II v |o V O Ph v |o Ph O o is s s o s o Pj v i! a| a| Ph A v S H 0h HH pq q pi Q Q s « Q a o < o < w 2 pa w < pa pa J w ►4 W J w i_4 pa Pa H pa pa o O O co ^ >d o -p -p H CU X> cu a o3 (S U Tt cu • TJ T3 -H 03 d c cu cu o o3 a3 co -p co p> O M •H •H -H o -P ■H CO fl a p a P •H 0) CU M O co CO O cu CU O CO 2 CO o 0) co CO -p •i-^ CU TH o Q co CCJ W ' N^^ o o O o ON t— t— t— ON C— t— t— t— £— t— t— t— t- t~ t— CQ O o o o o o o O o o o O O o o o o o CQ 0) A •— ^ *~-*s rH 1-3 J »-3 -P V^r * — ■ v_^ > pq Pn pq CQ pq ■a ' — ' — — ' — *—' H ON o t- ON r— t- t— C— t— t— t— t- t— t— t— t- t— O o o o O o o o o o o o o o o o o o CO a; * cu o J3 fl <«—-"» *-~"*> >» Cm Cm Cm CU a o — ' " — C\J GO o ON t— CO CO t— CO VO t— vo vo t— vo vo vo vo o o o o o o o o o o o o o o o o o o Jh o> cq CQ Pi fl fl y— s ft K i-3 CQ o « fl #* 43 Em Cm Cm to **— -" "■*--' *"■— ' co CO CM o vo r— t— CO vo vo vo t- t— c— vo vo vo t— vo o O o o o o o o o o O o o o o o o o "w CQ fi n * — -s CQ j CQ n fi .»~— N. #1 1" — K o o M O X — — — — — >> H Cm ON H o CO t- t— VO t— t— vo t— vo vo t- vo vo vo VO o O o o o o O o o o o o o o o o o o 0) -p 0) CO p 03 en EH EH T* a H CM on -3- H CM on _d- IT\ vo H CM on -3- H O d •H • fl o cd cd H H H H C\J CM CM CM CM CM on on a on -3" c5 PQ PL, a bO o d a) •H ^i p ft >> H ^i T) O o a CO co > a fl •H H 5n o P Cm O W co •H II II n 0) Cm O W P O a3 M •H cd ■a 43 •H P co d) M 43 43 < H 43 39 (i.e. its information is not used) would have its decay number incremented by 0.0005. On the other hand, if it were caused to output during that cycle, then the amount 0.0005 would be subtracted from the number. Thus if a T*-element after being "set" did not output for four cycles in a row, its output during the next read cycle, would have 0.0020, the decay number in this case , subtracted from the output value. Again, if the decay number for a T*-element at some point in time were 0.0025 and this T*- element is made to output its stored contents during the next two succes- sive read cycles and is not caused to output anything for the following cycle, the decay number after these last three read cycles would be 0.0020. This value (0.0020) would then be subtracted from the output of this T*- element if it outputs during the next cycle. Thus provisions are made for both a "loss" of memory from nonuse and "refreshing" of memory when information is repeated. A negative decay number is possible if a T*-element outputs for a sufficient number of time, however, this does not occur very frequently in an environment in which there are more than a few stimuli which are presented randomly. k.3 The Simulator The simulator is a PL/1 program. It consists of a main program FROG. Subroutines (or procedures) are written to simulate a general memory (so that under proper specification it simulates a GTM or a BTM or a PM), a MIU and a MIS. Function subroutines are written to simulate the logic elements S, E, P, T etc. Table U.2 lists the procedures and additional information about each. The source deck listing in Appendix B provides further explanations about those. ko Procedure Name Purpose How Invoked FROG calling program MIU simulates MIU MEMORY simulates a memory (GTM, BTM or PM) TSTAR simulates T*-element T simulates T-element MEM simulates MEM-element S simulates S-element E simulates E-element P simulates P-element SI simulates Si-element C simulates C-element MAX (*) simulates M-element MIN (*) simulates m-element MIS simulates MIS DECAY simulates decaying process of T*-elements COMPUTE simulates stepping process of T-elements RANDOM generates a random number on [0,l] for the random selection scheme main program "by CALL statement by function reference by CALL statement by function reference (*) Both MAX and MIN are built-in generic routines and, here, are not actual procedures within FROG. They are included in the list to make the list of program components complete. Table k.2 List of Procedures 1*1 The main procedure, FROG, includes the program to generate the stimuli in the various regions in a random sequence, and the simulation of the 'decision unit' (DU). The number of stimuli in FROG's enviornment is specified by- initializing the variable HOWMANY. The names of the various stimuli, their visual characteristics, their good taste values, and the pain inflicted by each one of them is contained in the arrays NAMES, CHARS, TASTES and PAIN. These are dimensioned at execution time by HOWMANY. The array CHARS has as many rows as the number of stimuli and four columns to accomodate the values of visual through visual ■ of the various stimuli. Thus CHARS (i,j) contain the jth visual characteristic of stimulus number i. The four attributes of the same stimulus occupy an identical place in the list, i.e. NAMES (2), CHARS (2,j) for j = 1, ..., U, TASTES (2) and PAIN (2) all refer to the same stimulus. At the start of the program the array RDPROB is initialized with values which represent the desired frequencies in the random selection scheme of appearance of each of the stimuli in the different regions. The number of rows n equals the number of different stimuli. The number of columns m equals the number of regions in FROG's environment. The arrays is initially dimensioned by the variables HOWMANY and REGIONNO which specify n and m respectively. The format of the array is as shown in Table U.3. k2 X 1 2 .... m I 1 P ll P 12 .... p i m P lx 2 P 21 P 21 • • n P nl p ran P nx I P xl P x2 P xm P Table U.3 Format of the Array RDPROB The quantity p. . represents the probability of occurrence of m stimulus i in region j. The partial sum P. = £ p.. represents the ix . =± ij total probability of occurrence of stimulus i over all the regions. The n partial sum P . = Z p. . represents the combined probability of occur- rence of the various stimuli in the region j . Obviously m m P=E I P. . = E P. = I P . = 1. i.l j-i U i-i lx j-l X J After the first executable pair of statements which read in the array RDPROB and prints it, the statement labelled BEGIN invokes the sub- routine RANDOM and assigns the generated random number to RANNUM. The following statement "IF COUNTER = n THEN G0 T0 TERM" terminates the pro- gram after n-1 trials. The next group of statements allow memory dumps at specified trials. The group of statements following the memory dump indicator uses the random number RANNUM to select which bug is to be presented to FROG. Let RANNUM = r. This block of statements obtains the smallest values of i and j which satisfies the inequality k3 i-i J . r < I P + S p k=l ta £=1 1£ These values of i and j specifies the stimulus to be chosen for the trial and the region in which it is to appear, and are assigned to the variables BUGCHOICE and REGCHOICE respectively. The variables BUGNAME, GOODTST and PAIN are assigned BUGCHOICEth members of the arrays NAMES, TASTES and PAININPT. The array VISUALCHAR is assigned the BUGCHOICEth row of the array CIIARS. The stimulus for the trial is now determined and its BADTST and INNATESTIM are evaluated. A message concerning the stimulus is now printed. The program now initializes all the memory outputs from each plane to zero and also the overall memory outputs. Read occurs from the memory plane corresponding to the region in which the stimulus has been presented by calling MIU which in turn calls the three memories (GTM, BTM nad PM) for that plane. Only if the NODECU output of this plane is 1, that the other planes are read. From the memory outputs the final outputs ALARMF, NODECF, MIUOUTF are determined. The value of the quantity FEED is also calculated. The statement following that simulates the "decision unit" (DU). This decides the course of action taken by FROG for the given stimulus. For each decision, the quantities HUNGER and FATIGUE are changed appropri- ately. The subroutine DECAY is called at the end to update the DECAYNOs. Appropriate messages are printed out in each case. If FROG decides to feed on the stimulus and no pain is inflicted by the stimulus on FROG, the visual and taste inputs are sent to the GTM and the BTM of the memory, for the plane, for recording purposes. The next statement is a call to ZERO- OUT, an entry point in the subroutine DECAY; this ensures that any kk T*-element that may have output during the write cycle will not affect the array USE. Specifically, the statement executed after ZEROOUT is called sets the whole array to binary zero after which control is returned to FROG. The COUNTER is then incremented and the control is transferred to the statement BEGIN to start a new cycle. k5 5. RESULTS AND CONCLUSION .06 .07 .07 .00 .08 .00 .07 .05 .09 .11 .00 .00 .00 .07 .00 .10 .00 .08 .00 .07 .00 .00 .10 .00 5.1 Summary of a Run of the Simulator The run consisted of 250 trials. The assigned probabilities of occurrences of the various stimuli in different regions are tabulated below. "^^Region ^\^ 1 2 3 k Stimulus F G H L B S The actual number of occurrences of the different stimuli in the dif- ferent regions were ^^Region ^^\ 1 2 3 k Stimulus F G H L B S So the actual frequencies were 2k 12 13 1+9 12 23 12 hi 27 33 60 19 21 Uo 23 12 3U 19 19 63 87 55 U5 250 H6 Region Stimulus - F .096 .0U8 .052 .000 G .0^8 .000 .089 .0U8 H .108 .132 .000 .000 L .000 .076 .000 .08U B .000 .089 .000 .0U8 S .000 .000 .076 .000 The frequency of good tasting bug . 38k , that of had tasting bugs was .1+00 and that of predators was . 2l6. A trial by trial summary of the run is given in Appendix A. 5.2 Results The reactions of FROG to the various stimuli are separately dis- cussed in the following paragraphs. Fly : A fly shares the visual combinations (1,2), (1,3) and (2,U) with a grasshopper, (1,2), (l,U) and (1,5) with a honeybee, (l,3) with a ladybug, (1,2), (l,3) and (2,1+) with a scorpion. Since flies are good tasting they were quite readily fed upon by FROG. Consequently they were learned fairly quickly compared to other stimuli. The first two times that a fly appeared in region 1, it was fed upon by FROG mostly out of curiosity and high level of hunger. This re- sulted in enough learning for it to show a memory directed approach in the third appearance (trial #23). In all the following encounters of a fly in region 1 FROG had no problems in correct recognition. In fact, a full learning was achieved by the end of trial #l6U (the 12th time a fly hi appeared in region l). Learning was reinforced because of grasshoppers in region 1. Recall, a fly shares a few level 1 and level 2 combinations with a grasshopper. In region 2 a memory directed approach was taken by FROG in the very first appearance of a fly (#20). This approach was based on what was learned about flies in region 1 in trial #s 1 and 3. Being tired FROG did not feed on it. But the same cause led FROG to feed on the flies which appeared in the trial #s 25 and 32. These feedings resulted in perceptible learning in the level h of GTM of region 2. The next time a fly appeared in region 2 (trial #75) FROG's decision was guided by the region 2 GTM output. Progressive feeding resulted in full learning after trial #135 (the 10th appearance of a fly in region 2). In region 3 flies appeared with grasshoppers again. Initial learning took place when FROG was driven by region 1 GTM outputs to feed on flies. The first two times that they appeared in region 3 (trial #s 28, 36). From the third time FROG was guided by its experience in this region (trial #50). The knowledge was pretty quickly reinforced in level 1 and level 2 by the simultaneous appearance of the grasshoppers with which flies share three combinations of basic visual characteristics. Full learning was observed after the 8th appearance of a fly in region 3 (i.e., after trial #126). It is to be noted that since the good tasting bugs are seldom avoided they are learned very fast. Therefore, these are hardly ever mistaken for anything else. In fact the problem is the other way round. The characteristics that some predators and bad tasting bugs share with some good tasting bugs cause FROG to make erroneous judgment about the former because GTM outputs for the shared visual combinations grow faster U8 to their full potential than the corresponding BTM or PM outputs. Grasshopper: A grasshopper shares the visual combinations (1,2), (1,3) and (2,U) with a fly, (1.2) with a honeybee, (1.3) with a ladybug, (1,2), (1,3) and (2,U) with a scorpion. It appears in region 1 with flies and honeybees , in region 3 with flies and scorpions and in region h with ladybugs and beetles. In region 1, following a few hunger forced reactions to grass- hoppers , FROG makes a favorable approach in the 5th appearance of region 1 (#71). The decision was mostly influenced by the GTM outputs correspon- ding to the characteristics shared with flies, which had appeared several times by this time. By the end of trial #250 grasshoppers were fed upon only seven times out of the twelve times that it was presented in region 1. So full learning was not achieved. But due to the shared characteristics between a fly and a grasshopper the GTM outputs were already quite close to the full potential value. In region 2 first few memory directed approaches were guided by the GTM outputs of level 1 (trial #39, u 5), and later by the level 2 GTM outputs of region 3 mostly due to the characteristics shared with a fly. At any rate, trends were similar to that observed in region 1 and full learning was demonstrated after the twelfth appearance of a grasshopper (trial #123) in this region. In region k no flies appear with grasshoppers. Initial reactions to grasshopper are thus guided first by curiosity, then by the GTM outputs of region 1 (mostly resulting from eating of flies) and then by the GTM outputs of region 3. In the sixth appearance of a grasshopper in region h k9 (trial #170) the experience in region k guide FROG to make its decisions. Grasshoppers were fully learned in region k after trial #189. The interesting thing to note here is that the characteristics shared by a grasshopper with a fly reinforced FROGs learning favorably for both a fly and a grasshopper. However, the same characteristics were shared by the scorpions. The reinforcement of the idea that these characteristics essentially relate to good tasting bugs caused FROG to suffer a lot when it came to dealing with scorpions as will be seen later. Honeybee : A honeybee shares the visual combinations (1,2), (1,10, (2,5) with a fly, (l,2) with a grasshopper, (l,2) with a scorpion. Honeybees appeared only in region 1 and 2. In region 1 honeybees appeared with both flies and grasshoppers Flies and grasshoppers are good tasting and are mostly eaten. Also the two together appear more frequently than honeybees. So for the shared combinations of visual characteristics development of an initial bias toward good taste is to be expected. The GTM output for the shared com- binations should grow faster than the corresponding BTM outputs, causing error in FROG's judgment. The results confirm the predicted trends. In the first few appearances of honeybees in region 1, they were mostly fed upon by FROG, directed solely by hunger. First memory directed approach occurred in trial #U8 based on what had been learned about honeybees in region 1. This knowledge was reflected in the avoid- ance of honeybees in the subsequent appearances (trial #51 s 59 > 60, 60). These decisions were made on triple joint combinations (level 3 50 combinations) since not much was learned about single and double joint combinations (level 1 and 2). But as the number of trials increased the level 2 and level 1 outputs become significant. Since honeybees share some of its level 1 and level 2 characteristics with flies and grass- hoppers, the reason stated earlier caused FROG to make some errors (#67, 73, 99, 117, 127, 150). Only after making six consecutive errors of judging honeybees as good tasting and feeding on them on five occas- sions FROG learned enough to recover from the spell as is evident in trial #151. Thereafter it made no mistakes with honeybees in region 1. At the end honeybees were not fully learned because they were mostly avoided. In region 2 honeybees appeared with flies only. In other words honeybees were not overwhelmingly outnumbered by the two good tasting bugs with which it shares some visual characteristics. So mistakes by FROG due to shared characteristics happen a little late (trial #96). With the exception of a minor sign of recovery in the next appearance (#105) the spell continued through the subsequent appearances of honey- bees (trial #12U , 132, 1U9) when they were fed upon (three out of four times) being taken as good tasting bugs. This feeding resulted in further learning so that FROG's decisions were corrected from trial #158 onwards. But due to frequent avoidance, honeybees were not fully learned in region 2 either. The behavior of FROG towards honeybees is a good example of how shared characteristics may confuse issues in the learning stage. Ladybug : A ladybug shares the visual combinations (1,3) with a fly (l,3) with a grasshopper 51 (1,1), (1,10., (2.3) with a beetle, (l,l) with a scorpion. It appears in region 2 along with flies and beetle and in region k with grasshoppers and beetles. In region 2 a memory directed approach is observed in the third appearance of a ladybug (trial #19) and they were avoided for a few sub- sequent appearances (trial #30, 56, 87, 90). As ladybugs were being avoided flies and grasshoppers were being fed upon. As the level 1 GTM output rose, the shared characteristic (l,3) with flies and grasshoppers inclined FROG to believe that ladybugs were probably good tasting too as is evident from FROG's reaction to ladybugs in trial #s 125, 1U1+, 152, 157, 159» Ladybugs were fed upon in these trials except when the hunger was too low (trial #125). By feeding on them FROG found out they were not so tasty after all so it started avoiding ladybugs again (trial #s 178, 19U, 212, 2U5, 2U8). In region U, for the first few appearances of ladybugs FROG was guided by its knowledge in other regions. The first time a ladybug was avoided, it was due to what was learned about it in the region 2. But ladybugs were fed upon the next two times when FROG was guided by the GTM output in region 3. These two feedings resulted in an appreciable learning in level k of BTM of region 1+ which kept FROG away from ladybugs in region h for quite a few subsequent appearances (trial #s 83, 85, 92, 98, 103, 106, 155, 156, 166, 169, 175, 18U). In the meanwhile grass- hoppers appearing in this region were mostly fed upon. By trial #21 U the GTM output for the shared characteristics (1,3) become more than 0.5 and dominate the decision. Thus ladybugs were taken as good tasting bugs in trial #s 2lU, 22U, 225, 250. These four feedings were not UNIVERSITY OF IliiNOTS AT URBAMA-CHAUPAJtK 52 sufficient in bringing the level 1 BTM output of region h up enough to offset the corresponding GTM output. This would have happened if full learning was achieved. Beetle : A beetle shares the visual combinations (1,1), (l,U) and (2,3) with a ladybug, (l,l) with a scorpion. It appears in regions 2 and k with ladybugs but no scorpions. In region 2 the first appearances of beetles were reacted to by FROG completely guided by its hunger. It knew nothing about them, so it had no particular basis for making a decision. Obviously, the attempts to feed on beetles were not very pleasant and the resultant learning caused FROG to correctly avoid them (trial #s 56, 122, 133, 1^0, lU6, 1U7) only mildly believing them to be predators. But in the same region lady- bugs also appeared with beetles. Because of the shared characteristics (l,l), (l,U), (2,3) between beetles and ladybugs FROG was slowly biased to believe the beetles as bad tasting bugs instead and to avoid them (trial #s 173, 183, 199, 205, 217, 232, 2Ul , 247, 2U9). Although this was not a correct diagnosis of the features of a beetle, it kept FROG away from the dangerous stimulus anyway. It was good in one way, but, on the other hand, FROG did not get to know a beetle properly. Of all the appearances of beetles in region U, only once, the very first time, did FROG have an encounter with a beetle. This time FROG's decision was initiated by its hunger. But in all subsequent appearances FROG was guided by its experience in region 2. For some time FROG avoided the beetles thinking them to be predators (trial #s 72, 131, 133, 136, 1U3, lU8), and rightly so, after which shared characteristics with ladybugs convinced FROG to believe that the beetles should be bad 53 tasting. Continuous avoidance caused almost no learning of this stimulus. Beetles were avoided but mostly for the characteristics they share with the bad tasting ladybugs. Scorpion : A scorpion shares the visual combinations (1,2), (1,3) and (2,1+) with a fly, (1,2), (1,3) and (2,U) with a grasshopper. Scorpions appeared only in region 3 along with flies and grass- hoppers. The first two times the curiosity of FROG led it into two un- pleasant encounters so that level k learning became appreciable to make it stay away from the scorpions in the subsequent appearances (trial #s 2U, 27, 33, 1+9). As the GTM level 1 and level 2 outputs rose with more and more feeding of flies and grasshoppers, the shared characteristics misled FROG to believe the scorpions, which appeared in the trial #s 53, 79, 8U, 9l+, 95, 107, 13l+ , to be good tasting. Resulting bad experiences (four times out of seven that FROG actually tried to eat on the scorpions) resulted in enough additional learning to make it stay away from scorpion for the next few appearances of them. Full learning was not achieved by the 250th trial. 5.3 Conclusion The results show that FROG was able to learn the various stimuli more or less correctly. However, there were mistakes made in the initial trials. There were three distinct type of errors made by FROG. The first was deciding to feed on a bad tasting bug or on a predator. These errors were easily corrected, for the feeding resulted in additional learning which alleviated the problem. The second type of error occurred when a good tasting bug was avoided. This error was difficult to correct; with feeding inhibited additional learning was not possible. Only starvation 5U forced feeding helped reverse the situation for this error. The third type of error was made whenever a predator was avoided, being taken as a bad tasting bug. Although this error inhibited proper learning, its effect was not altogether undesirable. In the case of beetles, even at the end of the run, (i.e. after 250 trials) they were not learned in the proper perspective and, therefore, problems similar to that resulting from type 2 error can be evidenced. Admittedly FROG is not perfect, it makes mistakes. But these errors are not undesirable. In fact, it makes the simulation more realistic and natural. 55 LIST OF REFERENCES 1. Lettvin, J. Y. , Maturana, H. R. , McCulloch, W. S. and Pitts, W. H. , What the Frog's Eye Tells the Frog's Brain , Proceedings of the IRE, November 1959, pp. 19^0-1951 • 2. Maturana, H. R. , Lettvin, J. Y. , McCulloch, W. S. and Pitts, W. H. , Anatomy and Physiology of Vision in the Frog (Ranapipiens ) , Jour, of Gen. Physiology, Vol. 1+3, No. 6, July I960, pp. 129- 176. 3. Marinos , P. E. , Fussy Logic and Its Application to Switching Systems , IEEE Transactions on Computers, Vol. C-l8, No. U, April 1969, pp. 3^3-31+8. 56 APPENDIX A TRIAL BY TRIAL SUMMARY OF A RUN OF THE SIMULATOR A trial by trial summary of the run reported is given in the following pages. A guide to the various columns is given below. Trial number - serial number of the trial. Stimulus - name of the stimulus presented in the trial concerned. Region - the region number in which the stimulus appeared. Location - the region number in which FROG was located at the time the stimulus was presented. Innate stimulus - value of the "innate stimulus" of the stimulus. Hunger Fatigue Miuout Alarm Feed Decision - current value of FROG's hunger. - current value of the signal FATIGUE. - value of the MIUOUTF signal for the trial. - value of the ALARMF signal. - value of the FEED signal. - mode of action decided upon by FROG for the stimulus presented. Abbreviations used are: F = FROG decided to feed. NF = FROG decided not to feed. T = FROG was too tired to move. P = FROG sensed a predator. F,R = FROG decided to feed on predator, and retreated to its safe resting place in region 1 on encounter, P,R = FROG sensed a predator too close to itself, so it decided to flee to its safe resting place. 57 Specific or Overall - A 'S' is put in this column if the decision was achieved from the memory outputs for one region and an '0' is placed if memory outputs from all the regions were used to make the decision. Primary reason - Entry to this column is the signal which was primarily responsible for the decision. Following abbreviations are used: h = hunger f = fatigue n GTM m = level m GTM output of region n. n BTM m = level m BTM output of region n. n PM m = level m PM output of region n. At times when the final decision was guided primarily by hunger or fatigue, the memory output is entered in this column along with the value of hunger or fatigue in order to show how well the cognitive memory worked. XTB.ISAO J.O oxjjosds 58 aS en a as •H » fl t- II II VO CO CO VO II II II CO II CO II [— CO CO • • • • • n II • • II • o J- j- o O o o -=r -3 J* o -3- o -a; o a a5 •H 0) II Eh EH II II II II 3 2 Eh g -H; 2 e II EH II s EH 2 II %* ca PQ O Pm O PQ Cm o PQ O Pm PL, X! CO H «M .C X! <*H on H CM CO H Jd CM A H CO A XI"BJSAO J-O OTJTOSdS o co O o O to CO O o CO o o CO O O CO o UOTSTOaQ fcSEH fefcEHCLi&HSPkPiHCuaCiH&HPM P83J o t— CO -a- LA LA CO o CO LA o o -H; LA t— CO o o LA o CO CO o CO -3" LA o o o t— o o o O o O o o o o o o o o o o o rajQiv ooo ooocooocoooooocoo OOO OOOVOOOVOOOOOOVOO q.nonT^ o Or—-* O O -* O _* t— OJ-Ot— O-3-OO OCOLA OOLAOLACOOLAOCOOLAOO ooo oooooooooo anStq.Bj OOO OOLALALALALAOOLAOOLAO VOOOVD J-LAVDJ-CMCOHOHHOHHO ooo oooooooooodddo OLAO LAOLAOLALAOLALALAOLALAO J39UTIU t— t— CO COCOCOONCTNCOONONOOt— CO MD ooo oooooooooooooo <3n ttittit n a OOO OOOOOOOOOOOOOO blL L llui . L +B o\ ON CO ON ON CO ONCO ON O\0O CO 0\ ON CO On On atyeuui ooo ddoooodooodddd UOXQ.'BOOT ^t r-\ r-{ HC\JHHH0JC\IC\JCOCOC0C\JC\JC\J uot9s^j CO C\J CM cvjojhcooj-h-cocococmcmojcoh snxnmiq.s co^tn wpqcucocn^co&Ho^w&HCOW •OfJ CO ON O HCMOO-d-LANOr— OOOnOHCMOO-J- H H CM CMC\JC\JC\JC\JCMC\JCMCM0OCOCOCO0O TBTJ.J, ooo oooooooooooooo 6o -H; -d" on LTN l/N LTN O o o LT\ O o o >> c NO II LTN II itn ll -=r >H o • • • • 01 en o -=T O _* o CM O s a S s S ■H d) II EH Eh II EH II u cc O o O cu A H 43 H XI H XJ <0 J- NO CM NO no NO u-\ ir\ LTN LT\ LT\ LTN LT\ on • • • • • • • no o o « o o « O o o o o II ll -3" II J- ll II II II n on -=t o on O CM on On on 2 Eh S II Eh II K EH Eh H -a- m e> pq O m pq pq Cm H CM XJ CM X! H CM CM CM on XT 6 j; sao - 10 OTjiosds CO CO co co co co co co UOTST3SQ Pn [X. Ex. [x, [x, fe Ix, Ex, fx, En En a fe a a fe a fc a a a a a Ph LTN pssj; no -d- l/N O o o LTN O -d- o o o LTN o LTN O -d- -3- -3- -d- O o o o ultbxv ° o o o o o o o o o o o o o o o o o o o o o o o o o o on no o q.noniw ° _3" LTN o o LTN d on LTN o o o o o itn o CM LTN -d- O -d" -d O O O ION l/N LTN LTN sriStq/ej <-\ h cm o h d d d d d o o o o o H O o o o o o o o o o o o o o LTN l/N L/N O O asSuriH no i/n _d- i/n _^ o o o o o ltn -d- o i/n ITS O LTN O LTN -d O LT\ LTN LTN O NO 1/N NO O O O O O snxnuiTq.s oo co co co oo s^vbuui CD o o o d o ON o On O on o CO o ON o CO o ON o ON o On o On uoiq.'BOO'x r-\ <-i on on h cm CM CM CM CM CM CM CM uoiSsy H on -=r h on CM CM CM CM CM on CM CM on snxnniT.q.g o En o En o pq k w En w o w co •ON ltn NO t— co On on on on on on X'BiJX, o o o o o o H cm on j- j- -3" J" -d" -3" o o o o o LTN NO fc— QO ON -d -d- J- -d- -d- O o o o o 6l -3- v£> VO CO MD on OJ -3" VO VD VO on VD on yo U"\ LT\ L/N Lf\ LT\ MD ITS IT\ LT\ lf\ LT\ vo ir\ \£> LT\ • o o o O O O o o O O o o O o o LT\ o s a II II II II t— II II II II II II II ii II II ii Jh o • II cd co -3" en on CM o m -d- cm -* on on on CM on on CT) cd 2 2 § 2 2 2 l>. 2 2 2 2 js; 2 2 2 on •H CD EH EH EH II EH EH EH Eh EH EH EH EH ^ B^ EH 2 !m K O pq PQ O pq pq O O pq pq pq O pq O pq Pi CM on eg CM on £1 CM OJ on CM CM CM CM j-\£>_3- o ooooooooooooooooo OOOOLrNLTNLTNOOOOOOOOOO atT.ST^.'Bj; OHOOr-tonHOHCMOOOrHOHO ooooooooooooooooo O LT\ O LT\ U~\ O O l/A •ISSUTIH t— MD VO t— t— CO CO 0\ CO t— t— OO 00 t— CO C— t— oooooood sriTnuiiaQ ooooooooooooooooo CO ONONO\0\0\0\COCO ON On ON CO ON CO ON ON o LTN O LTN o UN O O \o VO t— t— COO CO ON CO sq.'euu.i o uotq.'BCO r [ ojonononHHHHonoicMOJcMononojoj uotSsy onHOJoncMojojoncMHHHonojojcvioj sniranT q.g p^KPCcopqwjofeWWpciOKPLHpapq *Oft[OHCMon_cruNvot— cooNOHOjon_d-ir\vo Lr\Lr\Lf>Lr\L/NLrNLr\Lr\Lr\Lr\vov£ivovovovo^O "TBT.ZJ; OOOOOOOOOOOOOOOOO 62 LT\ UA on c\j CO on on on LT\ m on LP\ LT\ VO VD L/\ \D VO ^o vo VD LT\ VD vo • • • Lf\ • L/> • • • • • • • o o o • o • o o o o o o o o o o LTN o o ir\ o o £B II II II ii II VD II {T\ II l/\ II J- II -=r II IT\ II -3" II II • • • • • • g g g on i on g° ij o g o go §1° g O g^ •rl 0) EH H EH 2 Cu 2 EH II & II Eh II Eh II EH || Eh II EH II ^ K O o O 0- o (X. o o O b O o O (U CO on on C\J H OJ on x! on J3 rH A on x! H ,C on a on ^ IXBJgAO -TO o CO o CO CO o o o CO CO CO CO CO DIJTOQdS LA on O CM O o ir\ o O UA M3 O l/\ o \o ir\ LT\ J" UOTSTDSa r fct |x, fe UA o o P99j- LT\ LTN \D O l/\ O V£> IT\ IT\ J" -d" LT\ J" ooovoovoo o o o o o o nuexv ooolaou^o o o o o o o u-\ la on o cm o on on on on ir\ on on o o o o o o anSiq.'Bj h h cm on h cm o h cvj on h o h O IT\ LT\ O LTN LT\ O LTN O O IT\ O O •ISSUTIH CO f- \D VO V£> LT\ VO IT\ IT\ j- J" UA j- snTTiniTi.^ ooooooo o o o o o o On go On 0\ ao o\ On On ao ao ao oo On UA IA LA LT\ Lf\ C\J on H CM O sq/BUUi ooooooo uotq.'BOoq HrHon ^rj-HH h -d- h h h on uofSay; rHon^tajH-J-H J- h on h on on snxnuiTQ,g wfe^momw j c*< o tn o co °Nt— coosoHOJon j- lt\ vo t-- ao o\ T ^, T „ VO MO VO t— t- — t — t- — c— r— t— c— c— r— T^T^iI ooooooo o o o o o o 63 CM MO CM on CO on CO on CO on on LA on on LA LA LA mo MO MO MO MO MO MO MO LA MO MO O o o o O o O o O o o o o o LA o o LA o >> a II -3" II II n II LA II II MO n II LA ii II II a 1 LA Jh o • • • • • cd w B cd £JO on 2 1 4 £! o ^ d ° ^ £ o _* -a- pj -3- ho •H 0) Eh II S EH Eh H II H Eh II EH H II fcH fcH HH fcH EH II Jh k 5 O pq CO pq CO PQ CO PQ pq O pq CO (X. CM XX CM CM -=f CO A J- on x! CM on XI CM CM H -3- H x: XTBaaAo -to OT.jT.osdg CO CO CO co ro co CO CO ro CO CO CO co CO •Oil K fa fa fa A fa fa fa fa fa a 55 fa 55 fa S3 fa 55 fa a & fa 53 fa LA J- J- CM LA on o LA t— en o MO on LA LA on t— on LA LA t— on o LA o o o o o o o o o o o o o o UOJSJ09Q PSS£ o oooo oo oo ooooo TH.TBTV o oooo oo oo ooooo cm j-CMf-on t^ on ^on t— t— la o on Q.noriTi^ la j- la on mo onMO onMO ononLAHMO d oddo oo do ododd Q o oooo oo oo ooooo srujtq.'ej o OOHO CM O HO ooooo Q LA OLALAO LAO •I8lJtm.H _3- LA LA J- LA LA MO srixmnT^g o oooo oo oo ooooo CO ON CO ON 0\ 0\ CO On CO o\ > g II J- ll -=f II L/N ll _=r ll II U"N ll ltn ii j- II -3" ii II U"N II U o • • • • • • • • • 3 en a) §!° d ° CM o d ° -* S ^ d ° go d ° j- S d ° on 2 •H 0) EH II EH II En II E-t II EH EH II EH || Eh II Eh II EH EH II Eh ~ CG o C5 O o m O O O O pq 6 pq i, en x: 00 £1 cm ,c h ,c -j- H Si h .£ rH ,c on xi -=r H £1 CM oxjioadg w co CO CO CO CO CO CO co CO uoTsiosrr fe a a &H tn s &H fe Cm a a &H a Pn Pn a paa^ UA O LTN t— LTN o o ir\ t— ltn _d- -3- LT\ -=r en U~\ U\ -3- -3- on LTN J- • • • • • • • • • • • O o o o O o o o o O O o o o o o o o o o o o o o o o o o o o o o o o o en q.noni^ ^o d en no ltn on on o on no on no on no on no on d on no snST+vBj LTN o o o o o o H d o o o o o rH o o CM O o o o o o o r-i o j:a9uriH o o LTN o l/\ o o u^ o LTN Lf\ LTN LTN ITN J- _3" u^ LT\ ^J- • • « • • • • • O o o o o o O o snxnuixq.g o o o o O o O o o o o o » c n II u\ II no 1 LTN 1 J- II J- II u~\ II II II II II u o • • • • • • II II S CO cr! -3- go g° 00 o 2 g° g° £° on on on on 2 g d g ■H CD &H E-i II En II EH II Ei 1 EH II Eh II 2 Eh 2 Eh Eh Eh H Fh K PQ o O O O o O *■ — * pq cu PQ O O O 0. -3" on j3 on x on ,£ H X! H & on ja CM CVJ CM CM H CM H XXBjaAo .10 w w m m m w w en CO CO CO CO CO CO UOTSTOSQ £ fe fe fe £ £ Pn P-. Pn 25 Pl, Pn Pn Pn fc s t— LTN O O O U^ O O-3-O-d-OJ-UA p S3 j on lt\ vd ir\ -=f -3- m O-^-O-^-vou^^r do o d o o o ooooooo oo o o o o o ir\oi/AOOoo raj's TV oo o o o o o ltnoltnoooo od d d o o o ooooooo t— oo on on on on m oj-oj-ooj-on anonxw oo no no no no no no o .j- o -d- no u^ no oo o o o o o ooooooo oo o o ir\ o o uaoooooo 9n9TTej oo cm on on o o ooooohcm do d d d d d doddddo o lt\ o o o lt\ o ouAOiiAOLTNLPv J89unU U~\ W\ NO LTN J- H IA .J- -3" lf\ LTN NO ITN _3- OO O O o o o ooooooo qriTriTiiToc oo o o o itn o ooooooo L • +t> On ON CO CO CO J" CO ON ON On ON On CO CO SQ.BUUI do d d d d d ddddddd uoiq.BOO'T ,-irH h on on on on onon on on CM UOt39>J j- on on on on cm cm CM CM H CM sninrax^s j co o Pn pq a pq a a Pn o * °M vo t— 00 On O i*™ 3 3 2 2 d cm on -j- H H H H H H LTN \D t— CO On r-\ r-\ H r-\ r-i r-\ r-{ r-{ r-i r-{ 66 on on on LTN LT\ m en CM CM on VO VD ir\ MD ir\ LT\ \o \o C— t- vo • • LT\ • • • • • • • • O o • o o O o o o o o o o o o o L/\ o o LTN o o s C II LTN II J- II LT\ II J- 1 -=r II IT\ II -3- II J" II ir\ II -3- u o • • n • • • • • • • • si erf H O d ° on H O d ° rH O d ° go d ° g° d ° •H EH II EH II 2 Eh II EH II EH II EH II Eh II EH II EH II EH II fc « O O a, O O O O e> a b o P-. on jC on £ CM on ^ CM X! cm x: on xi on ^ on x: on .c H ,C .n-cjiosclg CO CO co CO CO CO co CO CO CO UOTST0SQ Pn a ex. P-H a Pn a Ph a Pn a Pn P^ pssj o o -3- o o o o o J- o o -3- O LT\ O -3" o uijbxv ° o o o LTN o o o o o o o o o o o o o o o o onoooonLTNUAononcMCMon ^noriTiAj \ovo ovo ltn itv vo vo r— t— vo oooododoooo OOOOOOOU^OOLTx snS t q. - ^ OHOOrHOooooo doooooooooo OOLT\OOl/NOOlT\0 jsSuhh ^ -* -* ir\ J- -a- uaj-_^- uaj- ooooooooooo sriTnuiTaq ooooooooooo •co oO On CO ON 0\ CO 0\ CO CO CO S0.-BUUT ........... *- O O OO o o o o o o o UOTQ/BOO'T CM _d- on on on on a uot9sh _=r on CM on CM CM on on on srr[riiiiTq.g o o « o P>H fc P«H o •on o P3T-ZJ. H CM CM on J- UA VO t- co On o CM CM CM CM CM CM CM CM on H H H H rH H H H H 67 la 00 oo on on on on LA LT\ LA VO VO o CM VO CM ir\ VO VO la VO LT\ • LA ■ • LT\ VO • VO LA • • la • . o • O o • • o • • o o . o O o o o LT\ o O LT\ o o LA LA o LA tf c II L/\ II LT\ II LA II LA II -3- II -=T II LA o II • II • • II II • II II * • II 03 U3 on d ° 00 3° g° on oo g° on m d ° d ° on i^ •H 0) 2 EH II 2 EH II EH II S 2 Eh II a S EH II EH II S En II U « M4 O Pm O O Pm PL, O PL) Ph a O Pm 3 PL, CM C\J Xi CM 0O -C OJ Xi CO on h x: on CM on Xi on xi CM CM Xi ITB.iaA0 jo 0TJT09dg o en coca cq ocow cqcoco w ra UOXSX09Q « fLt PL, fo ft fc fe PLiOnfe PUPufc Js; pL.Cn psajl o o o o LA o o LTN o o LTN LT\ o LTN o LA o LA LTN o o LA o o LTN -3- o LA LA O LA O o LA CM O CM LA O O LA o LTN o LA O o LA VO o VO LA O O LA o uia^xv ^nonx^ anSx^-Bj: aaSunjj snirvaiiq.g oo oo o ooo ooo o oo OsO\ ON ON CO OnOsOO O\O\C0 go 0\0\ o la ooo on ooon ooon on oon O LA o vo vo o o vo o o vo vo ovo oo oo o ooo dod o do oo oo o ooo ooo o oo OO H CM -3- Lr\00iH CMOO H OO OO OO O OOO OOO O OO LA o LA o LA LA O LA LTN O la LA O LA -d" LTN -=r la LTN -=r LA LA -3" la LA J- LA LA 3q.BUUJ OO OO O OOO OOO O OO uoxq-BOOT moo cmh h cmcmcm hhh on onon uoxSey j-cm CMon cm _d- oo h oncMon on _d- cm STrrrunx^g aq w pqco fc pqcoo copqo o pq j °Mhcm on.=r la vo t— oo o\ o h cm oo -3- -,-_ T T _ en on onon on ononon en j- j- ^j- -=f _* I^T^lL h h hh h hhh hhh h hh 68 on ononononooonononon VO LA lf\ ia VD VO vo \D V£> VO MD \D \D \o • LA LT\ LA • • • • • • • • • • o O O • o d o LA o O o O o LA o o o O o >» c II U"\ II LA II LA II II L^ II -3- II II II II II • II II II • • 1 • £ o on on CO §j O §j o i! 2* o go 2 1 2 1 £ CM 2 •H CO FH II 2 2 2 H II H II EH H II Eh II EH B^ E^ EH R Fh CC o CL, ex. ex. o O m O O pq pq pq O pq Oh iH 43 CM CM CM CM si H £1 H CM X! H 42 H _a- -=r CM CM X^BvISAO J-O OTjxosdg CO CO CO O CO CO CO CO CO CO co CO CO CO En pl, Pi Cm &H Pn q.noni^i 9nSTq.-Bj as9uriH aq.Buui o om En &H Pn En Ix, En 3 En S S s s En a UOISTD3Q P8SJ ,„.__, O LALALAO O OO O OOOOO UUB XV O LALAL/\0 O OO O OOOOO o o o o o o o LA LA o LA on o LA LA t— on on on o o o o o o o o o o o o o o on oooon on t-- on m t— t— t— co t— \o ooom3 v£> on vo ^o on on on vo on d oodd o oo d dodod o oooo o oo o OOOOLA <-\ CMOOO H CMO H OOOOO O OLAOLA O LAO LA OLAOLT\0 LA -^■-^•LALA LA -3- LA _4 U^LAVOVOVO O O sn T nui TQ-S o oooo o oo o ooooo CO ON ON ON 0\ 0\ 0\ , C II ii vo II II il ir\ il -a- II II II II II II II II u o II • • a m ti i!° CM H H O H O -=T -3- CM ^t OO -=r (\l CM a cd 2E, CO s; s s S E s 2 S 3 >l s •H co E-i EH II EH s EH EH II EH II En EH EH En EH Eh En Eh Fh CC Cj O m fa O O O m pq pq pq O PQ PQ pq Oh CM oo & H 00 CO H X! on 42 -3" -3- 0J _H/ -=f -=r H CM XIBaaAO ao co co co co co CO CO CO CO CO CO CO CO co UOJSTOaQ fa fa fa Ph fa fa fa fa fa fa fa fa fa a S a a a fa a a fa oo o c— o o o o t^t— r— f— l/nc— f— r- pggj VDVD OO O VO U"N J" OOOnoOOOU-NOnoOPO o d oooo o oooooood OO OOOOO o oooooooo UU'BXV OO OVDOO o oooooooo do dooo o oooooooo OOCM f— O 00 OO OO f — t — t^ t — LT^ t — t- — t — q.nOTiXM vot*- oo o vo vo vo onooooooLrvooooon do* dodo d oooooooo OlT\ LfAOOO o oooooooo anSxq.'Bj; oo hohcm co hoooohoo oo oooo o oooooooo IT\ O OUAOO O if\OLT\OU~\LT\Oir\ Jta9unH vo vo ua u-\ vo ir\ ^t _d-ir\Lr\vovoir\vovo oo oooo o oooooooo snTnnn: q.q oo oooo o oooooooo •°0\CO G\ C\ CO CO CO O\0\0\0\CO0\a\O\ STBUUT • .... . + J -oo oooo o oooooooo uotq.'BDoq cm cm on on uoxSsy; cm on on oo -J- -H; CM -3- -=r CM snxraniq.g ,_q ^ co O fa o J J a ^ o pq •on o> o H CM oo -3- vo VO vo vo H H H H VO VO t- vo vo CO VO On o vo t— CM oo TO CM t— >» a ii ^ o •H (U Eh Pl, on on vo on VO o o ii n CM CM Eh E-i pq PQ CM CM d II on S EH O on VO o ii EH PQ CM 00 VO s Eh PQ on vo EH pq CM ITN CO ■5* EH O CM EH O on on vo o ii En pq CM on VO -3- En pq CO VD d II CM pq CM en EH pq CM on MO o i Eh O on on VO o ii EH PQ LTN O II EH O OTJTDSdg CO CO CO CO CO CO CO CO CO CO CO CO CO U0TSTD9Q fc Cm Cm a Cm Cm Cm Cm a Cm Cm Cm Cm a Cm a Cm Cm C^ Cm o t— c— ua r— t- h- ia ia i- t—r— t— on t— ir\ P99J t— onontrNonononuNvo on on on on vo on irv oooooooooooooooo oooooooooooooooo ULIBTW oooooooooooooooo oooooooooooooooo cm t— t— ltn t— c— c— ir\ cm t— t— t— t— on t— u\ anoniM t^-ononuAonononu~\t"-ononononvo oolt\ oooooooooooooooo OOOOOOOOOLTNOOOOL^O SnSia'BJ' OHOOHOOOOHOOOOOO oooooooooooooooo O O l/AOO LTNOLT\VOI-rNOI-r\0 LIA LTN O jaSunu t~- vo vo c— \o vo t~- t— u-n ltn vo vo t— t— vo c— odddoddooodooooo <5TlTnmno OOOOOOOOOOOOOOOO L.+ooO On On CO ON ON On CO CO ON ON On On CO On CO 90/euui OOOOOOOOOOOOOOOO uoiq.i2Doq j- en on m on en m uoxSsy on CM CM on CM CM on H sn-[nuija,s Cm pq o pq o Cm pq pq o a o •°N -H- vo t— H CO t— On o H CM on -=f LTN vo D— CO ON CO CO CO CO CO CO CO CO CO CO H H H H H H H H H H 71 on CM OJ on on on on on 00 on on on CM vo t— C— vo VO VO VO OJ VO VO VO vo vo t— • • • • • • • VO • • • • • • o o ltn o o o o o o o o o o o o >> a ii II VO II VO II II II II II II II n ii II VO u o • • II • ^ H O g° §J CM ^ si OJ OJ CM §j si CM 2 d ° •H 0) H EH II EH II EH En EH EH s Eh EH Eh Eh Eh EH II U K PQ o O O pq pq O D-, pq pq O pq pq O cu H H 43 H 43 OJ CM H H on OJ OJ H Oi CM H 43 IIBJ9A0 JO co CO OTjiosdg en CO CO CO CO CO CO CO CO CO CO CO UOISTOSQ pn &H ps fo Pn fe #- s s a &H p* Pn Pn En Pn fe t^LTN ltn c—f— t— O O t— r— on t-— f— lt\ _ m vo ltn onononvooononvoononvo P39jI • • oo o ooooooooooo OO O OOOOCMOOOOOO ____. oo o oooovooooooo oo o ooooooooooo t — c\j cm c— t— t— m o i— s in t- t—cvi -——-.„. <*"• t— t— ononon^ooononvoonont— q.nonxjAi . . ........... oo o ooooooooooo oo O LTNOOOOOOOu-\00 snSx^ H . °. °. H . °. °.' °. H . «l *l °. °. °. °. oo o ooooooooooo O LT\ LT\ li^OLfAOOmu^UAU^OLTN t _q„ wtt VDVO U"\ J-L^U-\.Ol/Nir\l/NVOl/"NVOVO J.d3Ull[J . . a ... ........ OO o ooooooooooo ________ oo o ooooooooooo sn I nxu T-(-S on co oo ovo\o\r>o\cr\0\coovo\ac> sq.'Buuj oo o ooooooooooo uoiq.-B0OT j- j- H H on H H uoiSay CM CM on on CM CM H H sn-[nuiTi,s ffi fe w w o CO pq pq O w •°M o H CM on -3- UA VO t- OO Ov O H OJ on ON On 0\ on ON ON a\ On O O o o rH H H H H H H H OJ OJ OJ CM 72 CM CM CM OO on CM on on CM Lf\ CM on C— VD CM t— VD vo t— VD VO t— itn t— vo • • VD • • • • • • • • ■ • O O • o O o o o o o o o o L^ o U"\ L^ LTN LTN LTN o >J a II ITv II II LTN II -* II II LTN II II II is\ II J" II LTN II H o • II • • • • • • 3 s d ° ^ m d ° go u go ^ i! d o d o d o CM ■H QJ Eh II Eh s EH II EH II Eh EH II EH Eh EH II EH II EH II EH In K b pq Oi O O pq b pq PQ b a O pq a, r-\ 43 CM on H 42 H 43 CM H 43 CM CM H X! rH 43 CM 43 CM H'B.ISAO JO oijiosclg CO CO CO co CO CO co co co co co co co UOTSTD3Q fe fe fe fe fe fe fo fe s s fe s s (X, s fe fe PSSJ ITN LTN on o o LTN LTN ltn J- on ltn ltn t— on t— en LTN O o o O d o o O d o o LTN on TUJ'BXV o o o o CM O O O O O O O O O o o o o o o o o o o o q.nonxw CM t— on o o CM t— on on CM t* — t— on on CM ITN ITN CM f— ir- on O o d o o d o d o O O O d o o o o o l/N o o o o o LT\ O o o o o o LTN o o o o H o LTN LTN o LTN LTN o LTN LTN o ltn UA O o aaSunn m -3" LTN ITN -3" LTN LTv -=r LTN LT\ -=* itn j- d o o o O o o o o d O o d sniranTa,s ^ 90,T3UUI ^ o o o o o o o o o o o o ON CT\ CO CO ON CO ON On CO ON CO ON uoiq-BOon; h rH CM uotSaa h cm on on cm CM CM CM CM snirai-c+.g fe pq co O fe fe m •on -^ IBTJi CM LTN VD t— CO ON O H CM on ^S; LTN VD o O O o o H H H H H rH rH CM CM CM CM CM CM CM CJ CM CM CM CM 73 on on on on on CM tfN LTN on on on u\ on VO vo VO vo CM vo t— ITS LTN vo vo vo LTN vo • • • • VO • • • • • • • • • o o o o o o LT\ o O o o LT\ o o o o o o o s c II II II II II LT\ II -J- II LTN II -4" II ii 11 VO II LT\ ii u o II • • • * • • C/5 CM CM gj ^ on S° go d ° i!° i! u d ° S° i! •H <1> EH £— t E-i E-i 2 En II EH II EH II EH II EH EH Srf II EH II H Jh K pq pq o pq Ph O O o O pq pq O O pq pti CM CM -=r rH on h ,c H 43 -=r 43 -a- X! CM rH on xi -J- 43 CM oxjxosdg CO CO CO CO CO CO CO CO co CO CO CO CO UOTST09Q £j En En En Cm En En En En 3 En 3 En 3 En E-H En 3 f— r — L/^ f — O LT\ LT\ O LT\ C— t~- O O t~— p99J en en ir\ m o ir\ -J- ir\ j- on on vo u-\ on ddoddo o o o ooo o o oooocmo o o o ooo o o HUTS TV OOOOVDO O O O OOO O O oooooo o o o ooo o o t— t— on t— o on cm ir\ us t— t— on i/a t~- anon™ ononvoonovo t— u~\ lt\ on on vo ltn en '•**•••••• • • • ••• • « oooooo o o o ooo o o oooooo o o o ooo o o anSi^/ej; ooohoo h o h ooo h cm oooooo o o o ooo d o LT\OLT\Lr\OLr\ u-\ o ltn o its o O LTN JT99UT1H -J IA IA jj LT\ 1A _4" LTN _3" LT\LT\VO LTN -3" oooooo o o o doo o o snTUinxiQ oooooo o o o ooo o o '-•"^OnOncoOnOnoO co On On OnOnoO On On + x OOOOOO O O O OOO O O uoiq/eooq; cm cm cm _=i- UOlS9y CM CM -4" H on CM on snxnuiiq.g pq a o « co o a pq •on t- OO ON o H H H CM XBUJ, CM CM CM CM H CM on -=r in VO t— 00 ON O CM CM CM CM CM CM CM CM CM on CM CM CM CM CM CM CM CM CM CM 7U on on on co on on on on CM on on on CM CM VD VD VD vo VO vo vo VO t— VO vo VO t— t— o o O o o o o o O o o o o o o O O >» c II II ll n ll ii II II LT\ II J" ll ll II VO II IA II -d" Jh o • • • • • s CM g CM 2 H g CM g d ° d ° g s! d ° d ° d ° •H > CS il II LT\ ii II II II h o • s ^ d ° ^ 1 CM g ■H FEED THEN DO; PUT EOITCSTATE OF ALARM =•, ALARMF, •IMPULSE TO FEED = , ,MIUOUTF, •FROG HAS SEEN A PRPOATOR.') (SKI= FATIGUE THEN DO; PUT EDIT(« IMPULSE OF "FROG" TO FEED =' t F c EP, •FATIGUE LEVEL = • , FATIGUE, •••FROG" DECIDES TO FEED', •PFSULTING GOOD TASTF =',GPODTST, ♦PFSULTING eAD TASTF =',BAPTST, •PAIN INFLICTED =• ,PAIN, •FROG WILL NOW RECORD THE RESULTS OF THIS TRIAL IN ITS MEMORIES') (SKIP(2).X(<,0),A,F(10,7),SKIP,XUB),A,F(10,7),SK!P, X(49) ,A,3ISKIP,X(45),A,F(L0,7) ),SKIP,X(28) ,A) ; CALL OECAY; IF REGION=LOCATION THEN FAT I GUE=MIN( FAT I GUE + O. 050000, ONF ) ; FLSE DO; LOCATI0N=PEGION; FATIGUE=MIN(FATIGUE+0.1000000,ONF ); END; /♦WRITE INFORMATION OF THIS TRIAL INTO "EMORIES */ IF PAIN=0.0 THEN DO; CALL MEMORY (GTM , VI SUA LC HAP , GODOT ST ,GTMEMOUT , ZERO INT , REG tPN ) ; CALL MEMORYIBTM, VISUALCHAP ,BACTST, BTMFMPUT, ZFR 01 VT , PEGir>* ) ; CALL ZEROOUT; PUT EOIT('"FROG"S MEMCPIES HAVE BEEN UPDATED', •AND "FROG" IS READY FOP ANOTHER STIMULUS', STARLINE, STARLINE) (SKTP(3),XU0),A,SKIP,XU0),A,SKIP,A,SKIP,A); IF GOODTST>=BADTST THFN HUNGER=MAX ( HUNGFR-O. 1030000 , ZERO) ; 82 ELSE HUNGER =M AX ( HUNG ER-O. 05 00000, ZFRO) ; END; ELSE on; CALL MEMORY(PM t vlSUALCHAR,PAIN f PMEMnUT,ZER"INT,REr,nNI ; CALL ZFROOUT; PUT EDIT(«"FPOG" HAS ENCOUNTERED A PREDATOR SO IT», •RETREATS TO ITS SAFE BESTING PLACE IN REGION 1', STARLINE,STARLINE» (SKIP(3) ,X(37),A,SKIP,X(37) ,A,SKIP(2) ,A,SKIP,A); L0CATION=l; HIJNGFR = MIN(HUNGER+0.0 50C000 f ONE) ; FATIGUE = MIN(FATIGUE«-0.1000000,ONF) ; END; COUNTER=COUNTE"+l; GO TO REG IN; END; /* DECISION 5 */ ELSE 00; PUT C DIT( • IMPULSE OF "FROG" TO FEED ««,FEED, •FATIGUE LEVEL =•, FATIGUE, •"FROG" IS TOO TIRED TO MOVE • , STARL INE, ST ARL INE ) (SKIP(2),X(40),A,F(10,7I,SKIP,XUO),A,F(10,7), SKIP,X(40) ,A,SKIP<2) ,A,SKIP,A); FATIGUF = MAX( FAT I GUE-O. 2000000, 7.E RO ) ; CnUNTEP = COUNTFR«-l; HUNGE V=M I N (HUN GFR +0.05 00000, ONE) ; CALL DECAY; GO TO BEGIN; END; MID: PROCEDURE! VISIN, GTSTIN, BTSTIN, PAININ, mreGN, MALARM, MCUT, MNODEC, MPPNT ); /* SIMULATION OF A MIU. THIS SUBROUTINE IS INVCKED BY A CALL STATEMENT. THP PARAMETFP "VISIN" IS A FOUR-MEMBER ARRAY WHICH CONTAINS THE VISUAL CHARACTERISTICS OF TH= STIMULUS; ITS MEMBERS HAVE ATTPIBUTES FIXED DECIMAL (9,7). "GTSTIN", "AUSTIN", AND "PAININ" ARE THE VALUES ^F THE GOOD TASTE OP, THE BAD TASTE OF AND THE PAIN INFLICTED PY THE STIMULUS; EACH OF THE PARA- METERS HAVE THE ATTPIBLTES MXED DECIMAL (9,7). "MREGN" SPE- CIFIES THE RFGION IN WHICH THE STIMULUS HAS BFEN PRESENTED; IT HAS ATTRIBUTES BINARY FIXFD. "MALARM", "MOUT" AND "MNODEC" ARE OUTPUTS ALARMOUT,MIUOUT AND N0DEC4 OF THE Miu; THEIR ATT- RIBUTES ARE FIXED DECIMAL (9,7). "MPPINT" IS THE PARAMETER WHICH TELLS THE "MIU" IF A MEMORY DUMP HAS BEEN REQUESTED; IT HAS ATTRIBUTES FIXED DFCIMAL(l). IF "MPRNT"*1 A DUMP IS SUPPLIED; ELSE IF "MPRNT"=0, THEN THERE IS NO DUMP. */ DCL !VISIN!4),GTSTIN,BTSTIN,PAININ,MALAPM,GTMISIN!4),BTMISIN(4) , PMIS!N(4),MOUT,MN0DEC,MIS0UT(4) , ALAPMI SOUT (4 ) ♦ NOPEC IS T 0N( 4) ) 83 DECIMAL FIXED (9,71, MREGM BINARY FIXED, MPRNT DECIMAL FIXED! 1) ; CALL MEMORY(GTM,VISIN,GTSTIN,GTMEMOUT, MPRNT, MREGN) ; CALL MEMORYC BTM,VlS I N,RTST IN, BTMEMOUT, MPRNT, MREGN); CALL MEMORYIPM , VI S I N, PAIN IN, PME MDUT, MPRNT ,MR EGN ) ; GTMISIN(l)«GTMFMOUT( I); BTMISIN(l)»BTMEMOUT( 1); PMISINU) «PMEMOUT(l); CALL MIS(GTMTSIN(l),RTMISIN(l),PMISIN(l),MISnuT(l), ALARMISDUT( l),NODECISION( 1) ); GTMISTN(2)«SINODECISION(l) ,GTMFMOUT(2l ); RTMISIN(2)=S(NODECISTON(l) , BTMEMOUT < 2 ) ); PMISIN12) =S(NOOECISION(l) ,PMEMO|jT(2) ) ; CALL MIS(GTM!SIN(2),RTMISINC2) ,PMI S IN< 2 ) , MI SOUT (2 ) , ALARMISOUT(?),NOOECISI0N<2) ); N30ECISION(2)*MIN(NOCECISION( I ) , NOD EC I S ION ( 2 ) ) ; GTMISTN(3)=S(N0DECISION(2),GTMEM0UTf 3) I; BTMISIN(3)=S) ); RFTURN; END MIS; MEMORY: PROCEOURECCODE, VISUAL, NONVISIN,MEMOUT,PRNT,REGN) ; /* SIMULATION OF A MEMORY; THAT IS, A GTM OR A BTM OR A PM. THIS SUBROUTINE IS INVOKED BY A CALL STATEMENT. THE PARAMETER "CODE" IS A CODE WHICH SPECIFIES THE TYPE OF MEMORY. "C0DE"=1 REFERS TO A GTM. "C0DE"«2 REFERS TO A BTM. "C0DE"«3 REFERS TO A PM. THE ATTRIBUTES OF "CODE" ARE FIXED BINARY. THF PARAMETER "VISUAL" IS A FOUR-MEMBER ARRAY WHICH CONTAINS THE VISUAL CHARACTERISTICS OF THE STIMULUS BEING INPUT TO THE MEMORY; IT HAS ATTRIBUTES FIXED DECIMAL (9,7). "MEMOUT" IS A FOUR-MEMBER ARRAY WHICH IS USED TO RETURN THE 85 OUTPUTS OF THE FOUR LEVLS; ITS MEMBERS ALSO HAVE ATTRIBUTES FIXED DFCTMAL (9,7). "PRNT" IS THE PARAMETER WHICH TELLS "MEMORY" IF A DUMP HAS BEEN REQUESTED; IT HAS ATTRIBUTES FIXED DECIMAL (II. IF "PRNT--1, A DUMP IS SUPPLIED; ELSE IF «PRNT"«0, THEN THERE IS NO DUMP. "REGN" SPECIFIES THE REGION IN WHICH THE STIMULUS HAS APPEAREO; ITS PARAMETERS ARE FIXED BINARY. THE SECTION OF CODE SIMULATING LEVEL 1 IS DOCUMENTED; THE OTHER LFVFLS ARE SIMULATED IN A SIMILAR FASHION. IN OR THE F COMBI COMB I COMBI COMBI COMBI COMBI COMBI COMBI COMBI COMBI COMBI COMBI COMBI COMBI d;r to OLLOWI NATION NATION NATION NATION NATION NATION NATION NATION NATtON NATION NATION NATION NATION NATION COMBINATION UND = NG AR LEV l»VI 2»VI 3-VI 4«VI LEV 1»VI 2»VI 3«VI 4»VI 5-VI 6«VI LFVE l-Vl 2«VI 3»VI 4-VI LEVE l»VI FOR THE FOUR LEVELS: RSTAND THE MEMORY OUMPS: E THE CCMBINATORIAL INPUTS EL I SUAL(l) SUALI2) SUALI3I SUAL(4) EL 2 SUAL(l) .' SUAL( 1} ,' SUAL(l) ,' SUAL(2),< SUAL(2),' SUAL<3»,< L 3 SUAL(l) t' SUAL(l) f \ SUAL(l) f ' SUAL(2) »' L 4 SUALI1) ,VISUALI2),VISUAL(3),VISUALUI */ ,VISUAL(2) ,VISUAL(3) ,VISUAL<4) ,VISUALC3) .VISUALU) ,VISUAL(4) ,VISUALI2),VISUAL(3) ,VISUALI2),VISUAL<4) -VISUAL (3), VISUAL(4) ,VISUAL(3) ,VISUAL(t) /* /* /* /* DCL CODE FIXED BINARY, /* SPACIFIES WHETHER A GTM, A BTM OP A PM IS REFERRED TO */ THE FOUR BASIC VISUAL CHARACTERISTICS */ INPUT C ANO OUTPUT E FOR T*-ELEMENTS */ THE 4-TUPLE OUTPUT F FOR T*-ELEMENTS */ THE OUTPUTS OF ALL THE LEVELS */ INITM 16) O.OOOOOOO), THE INPUTS TO THE FOUR CHAINS IN LEVEL I */ INITM 24) O.OOOOOOO), THE INPUTS TO THE SIX CHAINS IN LEVEL 2 */ INITM 16) 0.0000000), THE INPUTS TO THE FOUR CHAINS IN LEVEL 3 */ INITM h) 0.0000000), THE INPUTS TO THE CHAIN IN LEVEL 4 */ /* THE ABOVE VARIABLES ARE SO DIMENSIONED THAT IT IS POSSIBLE TO SPECIFY FOUR INPUTS TO EACH T*-ELEMENTS IN EACH LEVEL. SINCE A LEVEL N T*-ELEMENT NEED N INPUTS (N=l,2,3 OR 4), SOME VARIABLES ARE NOT USED */ LEVIOUT, /* THE OUTPUT OF LEVEL 1 LEV20UT, /* THE OUTPUT OF LEVEL LEV30UT, /* THE OUTPUT OF LEVEL LEV40UT /* THE OUTPUT OF LEVEL PRNT /* PARAMETER WHCH TELLS (VISUALU), NONVISIN, MEM0UTI4), OUTPUTU), LIVISINPT(4,4) /* L2VISINPT(6,4) /* L3VISINPTU,4) /* L4VISINPTC4) /* 2 3 4 IF */ */ */ */ A RFGN ) DECIMAL FIXFD(9,7), MEMORY DUMP IS REOUESTED */ FIXED DECIMAL (1), /* THE NUMBER OF THE REGION IN WHICH THE STIMULUS 86 HAS APPEAREC */ FIXED BINARY, /* THE FOLLOWING TWO VARIABLES ARE ASSOCIATED WITH THE DECAY FUNCTION */ TEMPO) /* SCRATCH AREA •/ FIXED BINARY, CNTR(3) /* TEMPORARY MARKER, ONE FOR EACH TYPE OF MEMORY, TO SEE HOW MANY T*-ELEMENTS IN A CHAIN HAVE BEEN WRITTEN 1NTC */ 8IT (1), /* THE FOLLOWING FOUR VARIABLES ARE FOR TEMPORARY STORAGE OF THE OUTPUT STATUS OF EACH T*-ELEMENT IN THE BTM IN IN CASE A DUMP IS PEOUIRED */ (LAST0UT1(4,3,4,3) INITU144) 0,00000001, LAST0UT2(4,3,6,3) INIT((216) 0.0000000), LASTnuT3(4,3,4,3) INITU144) 0.0000000), LAST0UT4<4, 3,3) INITU3M 0.0000000)) STATIC DECIMAL FIXEC 19,7), /* THE FOLLWING LIST OF STORED VALUES FOR ALL THE MEM-ELEMENTS AND THE STORED VALUE AND FRACTION MULTIPLIERS OF ALL THE T-ELEMENTS */ UL1TST0VAL(4,3,4,3),L3TST0VAL(4,3,4,3),L1TPFRACI4,3,4,3), L3TPPRAC(4,3,4,3)) INITIAL (1144) 0.0000000), (L1MEM(4,3,4,3,6),L3MEM(4,3,4,3,6)) INITIAL ((864) 0.0000000), (L2TST0VAL(4,3,6,3),L2TPF*AC(4,3,6,3)) INITIAL ((216) 0.0000000), L2MEM(4,3,6,3,6) INI TI AL (( 1296) 0.0000000), (L4TSTQVAL(4,3,3),L4TPFRAC(4,3,3) ) INITIAL((36) 0.0000000), L4MEM(4, 3,3,6) INITIAL I (2161 0.0000000)) STATIC DECIMAL FIXFD (9,7); /* LEVEL 1 */ L1VISINPT(*.1)=VISUAL; /* SET UP VISUAL INPUTS FOR EACH CHAIN */ OUTPUTU )«ZERO; /* INITIALIZE LEVEL 1 OUTPUT TO ZERO */ DO 1=1 TO 4; /* PARAMETER "I" IS THE NUMBER OF THE CHAIN */ NBHODD=ONE; /*TURN ON SIGNAL "NBHOOD" AT BEGINNING OF EACH CHAIN */ CNTR(CODE)»'0«B; /* INITIALIZE TEMPORARY MARKER */ DO J=l TO 3; /* PARAMETER "J" IS THE NUMBER OF THE T*-ELFMENT IN CHAIN "I" */ CALL TSTAR(NONVISIN,ONEINT,L1VISINPT(I,*), L1MEM(REGN,C0DE, I , J, * ) , L1TST0VAL (REGN,CODE, I , J) , LITPFPAC(REGN,C0DE,I,J),LEV10UT,NBH00D) ; IF LEV10UT-»ZER0 THEN USE1 ( REGN,CODE , I , J )« • I • R ; /* SET UP "USE" ARRAY WHICH WILL INOICATE WHICH T*-ELEMENTS OUTPUT SOME NONZERO VALUE DURING A READ CYCLE *7 ELSE GO TO Li; /* IF THIS T*-ELEMENT HAS NOT OUTPUT SOME NON-ZERO VALUE OURING A READ CYCLE */ 0UTPUT(1)-MAX(0UTPUT( 1) ,LEVIOUT-DECAYN01 (REGN,COOE , I , J) I J /* IF THIS T*-ELEMENT HAS OUTPUT SOME NON-ZERO VALUE, THEN DECAY IT BY THE CORRECT AMOUNT, AND XEEP A RUNNING MAXIMUM TO FIND THE LARGEST T*-ELEMENT OUTPUT FOR LEVEL 1 */ LI I LAST0UT1(REGN,C0DE, I , J ) -MAX( L1TST0VAL ( REGN.CODE, I , J ) ♦LlTPFRAC(REGN,COCE, I , J J-DEC AYN01 ( REGN,CODE , I , J ) , ZERO) J /♦ SET UP THE OUTPUT STATUS OF THIS T*-ELEMENT IN CASE A DUMP HAS BEEN REQUESTED */ IF -.CNTP(CODE) 6 L ITSTOVALI REGN,CODE , I , J)»ZERO THEN DO; /* WHEN FIND THE FIRST T*-ELEMENT THE CHAIN THAT HAS NOT BEEN WRITTEN INTO THEN SET "MEMCNT ( CODE) " EQUAL TO 87 THE NUMBER OF THE T*-ELEMENT IMMEDIATELY TO ITS LEFT. THIS NUMBER IJ-l) WILL TELL "DECAY" HOW MANY T+- ELEMENT IN EACH CHAIN HAVE BEEN WRITTEN INTO AND, HENCE, HOW MANY T*-ELEMENTS IN EACH ROW TO DECAY */ CNTR(CODE)»«l»B; /* ONCE A VALUE HAS BEEN SET FOR "MFMCNT(CODE)" IN THIS PASS THROUGH THE CHAIN THEN TURN ON THE SIGNAL THAT WILL PROHIBIT ENTRY INTO THIS DO-GROUP UNTIL THE NEXT ITERATION OF "I" */ TEMP(COD6)=J-l; MEMCNT ( CODF)*M AX (MEMCNT! CODE), TEMPI COOE) ); END; END; END; /* LEVEL? */ L2VISINPT! 1,1) = VISUALU) ; L2VI SINPT( 1,21 » VISUAL (2); L?VISINPT(2, I ) = VISUAL! 1) ; L2VISINPT(2,2)=VISUAL(3); L2VISINPT(3,1)=VISUAL!1) ; L2VISINPT!3,2)«VISUAL!4>; L2VISINPT(4,l)»VISUAL(2); L2VISINPTU,2)»VISUALC3); L2VISINPT(5,1)=VISUAL!2); L2VISINPT(5,2)*VISUAL(4); L2VISINPT(6,1)=VISUAL(3) ; L2VISINPT(6,2)=VISUAL(4); OUTPUT(2l=ZERO; DO 1=1 TO 6; NBHOOD=ONE ; DO J=l TO 3; CALL TSTAR(N1NVISIN,TW0INT,L2VISINPT{I,*), L2MEM(REGN,CODE, I , J,* ) ,L2TSTOVAH RE GN, CODE, I , J ) , L2TP c RAC(REGN,CODE, I , J I , LEV20UT , NBHOHOI ; IF LEV20UT -=ZFRO THEN USE2 ( REGN.CODE, I , J ) = • I • B ; ELSE GO TO L2; OUTPUT(2)=MAX(OUTPUTI 2) ,LFV20UT-DECAYN02( REGN,CODE , I , J ) ) ; L2: LASTOUT2(REGN,CODE, I , J )=MAX( L2TSTOVALI REGN ,CODE, I , J )* L2TPFRAC(REGN,CDDE,I,J)-OECAYN02(REGN,CODE,I,J),ZERO); END; END; /* LEVEL 3 */ L3VISINPTI 1,1)* VISUAL! I); L3VISINPT! 1,2) "VISUAL! 2); L3VISINPTC1,3)*VISUAL(3); L3VISINPT!2,1)*VISUALI1); L3VISINPTC 2,2)= VISUAL! 2); L3VISINPT! 2,3)= VISUAL!*); L3VISINPT(3,1)»VISUAL(1); 88 L3VISINPT(3,2)=VISUAL(3) ; L3VISINPT(3,3)=VISUAL(4) ; L3VISINPT(4,1)=VISUAL<2) ; L3VISINPT(<,,2) = VISUAL(3) ; L3VISINPT(4, 3)=VISUAL<4) ; 0UTPUT(3)=ZER0; on 1=1 TO 4; NBHOOD = ONE; DO J=l TO 3; CALL TSTAP(N1NVISIN,THREEINT,L3VIS INPTf I,*> , L3MEM(RPGN,COnE, I , J , * ) ,L3TST0VAL (RFGN,CODE, I ,J), L3 T PF P AC (REGN, CO DE,T, J ),LEV30UT, NBHOOD) ; IF LEV30UT -.= ZERO THEN USE3 ( REGN ,CODE ♦ I , J ) = • 1 • B ; ELSE GO TO L3; OUTPUT (3)=M AX (OUTPUT (3) ,L EV 3PUT-CFCAYN03( REGN , CODE , I, J) ) ; L3: LASTOUT3(RFGN,CODE,I,J)=MAX(L3TSTOVAL(REGN,COD=, I, J)* L3TPFRAC(OEGN,CODE, I , J) -DECAYN03 ( REGN, CODE, I, J) ,ZERO); END; END; /* LEVEL 4 */ L4VISINPT=VI SUAL; 0UTPUT(4)=ZER0; NRHOOD=ONF; DO 1=1 TO 3; CALL TSTARCNONVI SI N , FOUR I NT ,L4V I S INPT , L4MEM(REGN,CCDE,I,*),L4TST0VAL(REGN,C0DE, I) , L4TPPR AC (REGN, COPE, I ), LFV40UT, NBHOOD) ; IF LEV4PUT ^= ZERO THEN USE4( REGN .COD^ , I ) = • 1 • B; ELSE GO TO L4; OUTPUT (A) =MAX(OUTPUT(A) ,L EV40UT-DPC AYN04( REGN, CODE, I ) ) ; L4: LASTOUTM REGN, CODE, I) =MAX ( L4T STOVAL( REGN, CODE , I ) * L4TPFP AC(REGN,COCE, I )-DEC AYN04( REGN, CODE , I), ZERO); END; /* SEND OUTPUTS OF ALL LEVELS OF MEMORY BACK TO "FROG" */ MEMOUT=OUTPUT; IF PRNT=1 THFN DO; DO K=l TO 4; IF CODE=l THEN OH; PUT EDIT(«nUTPUT STATUS OF EACH T* ELEMENT IN REGION «,K, • OF GOOD TASTE MEMORY IS NOW:»,»T* ELEMENT NUMBER* , •1' ,'2» t'3' ) (PAGE,SKIP(4),X(10) ,A f F ( I ) , A , SK IP ( 2 ) ,X( 64) ,A, SKIP,X(52),A,X( 19),A,X( 1<») , A, SKIP) ; END; ELSE IF CODE=2 THEN DO; PUT FDITCOUTPUT STATUS OF EACH T* ELEMENT IN REGION «,K, • OF BAD TASTE MEMORY IS NOW: , f , T* ELEMENT •1« ,'2» ,'3») (PAGE,SKIP(4),X( 1C),A,F(1) , A, SK IP( 2 ) , X( 64) , A, SKIP,X(52),A,X(19) ,A,X( 19), A, SKIP) ; END; ELSE DO; 89 PUT EDIT! 'OUTPUT STATUS OF EACH T* ELEMENT IN REGION »,K, • OF PREDATOR MEMORY IS NOW:«,»T* ELEMENT NUMBER •♦ •l«,»2 , ,*3 i l (PAGE»SKIP<*),X(10),A,F(l),A,SKIPm,X(64),A, SKIP, XI 52), A, XI 19), A, XI 19 I, A, SKIP); END; DO 1=1 TO 4; PUT EOITI'LEVEL 1» , ( LASTOUT1 IK ,CODE, I, J I DO J«l TO 3), •COMBINATION »,I) ISKIP= 0.5 THEN THE VALUE OF "GATE" IS RETURNED TO THF POINT OF INVOCATION; IF IT IS < 0.5 THEN ZERO IS RETURNED. */ DCL (SIGNAL, GATE, PFIVE IN IT ( 0.5000000 ) ) DECIMAL FIXED (9,7); IF SIGNAL ZERO, THEN THE VALUE 1.0 IS RETURNED TO THE POINT OF INVOCATION; IF "SIGNAL" = 0, THEN ZERO IS RETURNED. */ DCL SIGNAL DECIMAL FIXED (9,7); IF SIGNAL>ZERO THEN RETURN(ONE); ELSE RETURN(ZEPO); END si; C: PROCEOURE(NUMBER) DECIMAL FIXED (9,7); /* SIMULATION IF C-ELEMENT. THIS FUNCTION IS INVOKED BY THE USE OF "CUNPUT TO C-ELEMENT)" IN AN EXPRESSION. THE ARGUMENTS MUST HAVE THF ATTRIBUTES FIXED DECIMAL (9,7); THE FUNCTION RETURNS A VALUE WITH THE SAME ATTRIBUTES. THE PARAMETER "NUMBER" CORRESPONDS TO THE INPUT TO THE C-ELEMENT; THE CONTINUOUS LOGIC COMPLEMENT 1-INPUT IS RETURNED TO THE POINT OF INVOCATION. */ 93 OCL (NUMBER, ONE INIT ( 1 .OOOOOOO) ) FIXED DECIMAL (9,7); RETURN(ONF-NUMBER) ; ENO c; (NOFIXEDOVERFLOW) : RANDOM: PROCEDURE BINARY FLOAT (31); /* RANDOM IS SUBROUTINE THAT GENERATES A RANDOM NUMBER ON THE CLOSED INTERVAL BETWEEN ZERO AND ONE EACH TIME IT IS CALLED THE VALUE OF THIS RANDOM NUMBER IS THEN USED TO DECIDE WHICH STIMULUS IS TO BE PRESENTED TO FROG AND IN WHICH REGION. */ DCL (NUMBER, TEMP, FLCONST INITl I 111 11 llll 11 11 111 I ill 111 11 Ull 1F0B)) BINARY FLOAT (31) , (K STATIC INITI27), FIXCONST INIT(2l*7483647) ) FIXED BINARY (31); K*K*(2**l6-3); if k- -> — o o i/>« h- « 13 • X O o o OC o > u. o u« o c. • • o ac UJ o u. or o- I < 3 o e e o o e o» ♦ e © o • • « z o o o o e o O Q o o o o • — o o o o o o o o o o e 1- ac f> luepoe o ac oe o go O o o o o o o o — e a e u o. e o o IA e o U. • • t • O o (/l UJ 0* e o o o o o o DC o e O t- o o o o o oz t • • — o o o o • •• X «- o © O z o o o o o * 5 o o o e UJ z o O . • r • — 1— o o o o z o mm Z O Ul KM O O O O O — OC X X -» o o © o o u >- * 1- O O © O UJ OC ac •*• •» © O © O ac O « X z * ac o © © o u. UJ >- o z u> o o o o at «/> — MM o z Ul O © O O C 3 1 o o t- t . • . u. — O "-i UJ o — z o o o o z or Ul IS «" 3 Z UJ a UJ u of ac _J z a a CL — — — — -1 UJ U- o ac O Z Z Z Z U. 1- — mm IL O Z 3 3 3 3 O 3 jl ac V u. uj co cc ec cc q. Uj ac >■ Z 3 3 3 3 1- 1- < 0. O a V l/> l« 3 3 o x z O o u. 0. O o uj UJ z O — - ac z Uj z 3 Z u. >• z Ul o -J _i _j -i O ar ■ z UJ z 1— Ul Ul UJ UJ •* « IM. UJ 3 > > > > -J 1/1 t- ac a. uj uj uj uj «j •» in O ►— — i — i — i — i I— « i— 3 H- •J O Q o o t-5 X ui X > ZU t/l O U» z UJ 3 O D I 5 — UJ l- — 13 l/> > Ul 4 z uj ac » — or o. ac Z © UJ 2^gJ «/> Z u. O — UJ uw>-z Uj lu OC — O ac O a. r u O UJ 3 1-OItB 3 uiciui Z u — O uj ec X ul< >- I— UJ i- •/) X Qh l- Z 3 j» — OQ J o to z -I < 1.1 3 -J ui CulJu O OC -- J» UJ X UJ ■ co «- OThgc O uj — UJ a z a HIUOX ( ac i/i uj © o o o o e o o o mm o © © • z o e O rt o o o o o o o o — o a o o © 13 O © o o © ui o o § o o o © e ac O O o o o © o © o o © o o o e © o fc- • • o o f~ © e p* tm\ MM o o «M © © O i 0" o © 1/1 o o o o o f-l © o 1- o o O o © • • — o o o o o o o z© o o o o — 3 O O O O CO o o o o « z 3 O O O O — c l/> • • o o o o Z O z C uj o o o o o M- ac ■» o o o o (3 — t- o o o o UJ OC _J " o o o o ac c z ac o o o o u. o z o o o o o a MM o z uj O o o o C 3 o & 1— • U. MM UJ 13 MM z o o o o z oc UJ ac (3 UJ ■" 3 — U- ac OC o OC 0C m- LL o OC c z JT z 7 U. M- U. c X 3 3 3 3 O 3 V u. UJ 03 IE cc no a. ac > X 3 t5 3 -) *— M- o ac > trt ul oO ul 3 3 X u OC u. 0. c UJ X o o — f-J **N * t- z UJ z 3 X X UJ IrtJ -1 _J -I C OC UJ z 1- UJ UJ UJ UJ * 1- UJ 3> > > > _J */» OC a uj UJ UJ UJ « < t/> o 1— 1— •- 3 O X »1 t- 3 _J O U. 3 O © a x o o — o o l/> t- o o o u i« o in 3 MM u> im as i— uj 43 • V KTI .. I- . O — z < o 1 or O 2 _J UJ MM Z •• Ul •X o-t — «I J UJ Jisoc « a 3 — < c a uj 13 x u- uj o — 13 OZ Z Z h- 3 — uj : CO _J « ac ul 3 Ul — ul X X — t- l- >o UJ 3 < OIU. 98 o e e e e a o e e s e e o e e e e e e e o e e e e e e e o e e o o e e o e e o e e e e o o o * e e e VI o r> i»i o o e UJ o in m o o o *■ o o — o a * or o 00 •■■ e o • O r* o o -J O 1- » o »>- vi e o o o ^ o IN o fn O O O O o o o < UJ l/l i* M IN ►- o o o o •" •0 o IN !:§888*... o NO o o «— I- • — o o »■ o ■O O O o o or •« or o O a z o OOOSN o a O x o o o o o m 5 o o o o 30 • O Uj • e» o h- O uj so o o o o. O o o o • o a. X oc e o o o •• z c o o a © - z N o ir u. oo i/i 31- 30 O O O (M o 3 o o e e m o 3 1 1 IN H o Ma o i/» • t/> • • • • MM O O • B 1 o Z XX O O O O Z 13 © o o o z is e o UJ o h- UJ < Z O UJ Z O UJ C O O O O — or ^~ 1 UJ h- UJ • UJ C O O O O — CL H UJ N tA i^ t- O U. •ear — O O O O J — a o o o o Z 3 It Ui « 1/1 O o IN ►- o OO^u.* *i ^ireeeuii 3 -J o >- « » UJ U. in « o O O ••> or O m •* m o o o or □ X o uj — ►- l/l > z IN or o O O IN u s m eccooo u. Z 1- > o o Q >- •« >• o z o o o o *• at z o O 3 M o z Ui « o o O O 3 — o ■ -J o O « K 3 « o V c ►- • . • . U. — (9 MM o l- . . • t u. — M IS O 03 O W) U^ UJ UJ IS — ze o o o x UJ IS to* Z O O O O X < 1- O UJ I ■-« UJ uj or or eg »- T X IL r or o z Z Z Z U. t- IL O at O Z Z Z Z u. i- X 3 3 3 5 O 3 z o u. « I -t i 1* UJ I u. o X 3 3 3 3 O 3 Ik O MM C u. 3 z X IS > u. UJ OS SOffi Q. >■ IL uj os ob s3 a a. X !/> I/I m O o et > X 2 3311->. at > X 3 3 T 3 >- l- J» or UJ UJ Ui « or i/> or O a V l/l IAUIM33 O OC t> u» 3 3 or « tjO a or a o ■ u. X UJ o CL c 0. o X Ul Q X or a u. a. o □ S -J 5 UJ 8' x uj X 3 X Z UJ X 3 X UJ o. oc or o UJ X UJ X l/» -J *- UJ UJ UJ UJ « UJ X UJ X uij J J JOI >- UJ UJ UJ UJ < X X Z u. Z i_ UJ 3 > > > > -J 1— UJ 3 > > > > -J _/ O i/i Of a. uj UJ UJ UJ 4 l/l *- or a. uj uj uj uj < « Z •w t/i a •- -i -J -J J « l/» O >— i — i i — j z p— < h- 3 l»- «■ i— 3 MM ■J ■" O ►- « IL -j o o c o < as O UJ or CL o c B o o •* o UJ or a. z IS 99 i i Art o •- i/> o o o UJ s o *> ■■ ■* o •n S O r- e K O IM o M n z o « e * Uj o • z o o o o o Z o o o o o o o o o o o • — o o o o o o l/l -* ^ KEG 0000 OOOO 0000 0000 00 0000 ♦ ►- Irt m ■ l/l 1* s o C« o o o MM •J or u. a, o o o K o o z i U1M o o SO • O Uj • mm O v O Uj o OZ o UJ so e o o * o e o 5 •• x 5 O O O O r" c a- o OO u o • o a x <£ «■ K XI* » o IT U. O o i/i 31- or • s go II O B o o UiO o l/loP ■/!•••• — • O II o X z z or ar- K o e O O Z (9 ee >- UJ o ►- UJ « UJ ia 1/) Z O UJ 5 o o o o — ac I- ■ i*i ►- UJ UJ - o O o o uj ar i a U. UJ « u-> o o 5 o o^ m s — -j e ►- « • UJU. Z OZ o at m <«eee«c z Ou> — »- l/l > -1 3 — m o e «X UJI- 8 z CI 32888.^ Su. 1- > l_> UJ UJ 8 O o UJ *- J al >■ xo « > o i/i M mm c z u' >~ o o o c a — c ■ -j ci o « m- a < — <_) ►» X o mm o *-.... u. — N» ts O e <-> ia IA UJ a tw • 8 IU o zo o o o z « ►- Ouj ( UJ uiar t- zz o l/l or IU 19 — s or a at a O O cj — 1 ar ujuj Z IU or. UJ u> a. U 19 C Z z u. m. v> 10 u, vi u or. ■ >■ I- >- m- m. r o lt>- 1—01 »- kri z UJ O — •* 1/1 UJ _■ 2 a ar a ■-■-•— ^ m- S >- u. m- m- M X z ar o -J IU u. O ar O Z Z Z Z U. H- z o u.< • -i 1 *■ uj a i/> ►-a. o ■M M u. o z s s s a o s to* OIL a z z o *- t/» o a at V u. uacaa a. z i/i i/i MM o o X Si/l o IU s >- z s s s s ^ »- vac IU IU UJ al a i/i ar »- T — o ■ a. or V MMMVI33 at « l/l ar ar a O t u. o 8 K z o i u. a. O ii -J l_> Ul l/IB • UJ IU z a UJ o «o • ■ z IU at UJ a. ac aro IO o o u. V z UJ mj j j joc Z z u. z a o I z UJ z •- UJ UJ U-' UJ < M X ■ « ■ u- o « k HI a > > » > -j J a '-5« o i/» PB at a. uj Uj uj u. « al z a • • o « M a »— _J — ' _< — J z at IU o a» « — a _J IL or 00 K- < S u _l 1 m o s ■ p> • o i/i vi a 1 o o o o u IL o M o a) u, O lA _J o IU at a u. o i/i -J X 13 o ar u. IL o o IU IU UJ O o o > > > o o UJ IU IU l/l ^• o o _i _) _j U l/l o 0> o o u. u s N* UJ <* • O O «/» •• >- • o >•— Z « o l/l *> l/l JO. O Z •• *■> fc> N> U. UJ — Z .. t/> S S 3 *• *• IA " a. a. a. ■« o « mm — N k- « z -J IU g S o 3 O — « o or W IS X u. UJ e — C <-) z z Z 1- UJ 3 « Ij O X at or vi3ui "HI I - m- i-> O 100 U —I > a 3uj z r i z s o z o « u. a 3 uz o 3 — e a o *■ o « _l o- ^ Z Z o UJ u U (/» i/i u. a o i- a. o i/i o 3 l/> o ^ — o o I N. < O • I o o o a. o 1 a o O t o o ■ • e a UJ o u. a. o- — i/i XX i « X — O Zm ►- 3 _i e a. 3 © o o zoo — o o i/> >- o o is u o o 3 — ifN O a. >- UJ |/N • >■ f> •• t- • o o — z « o < « o z _| UJ — Z •• «/i t- »- — ■• « <_> ■* « < X -1 UJ uo a. a ««3 — •< C at uj O X u. UJ o — ouz 7 ?"- u 3« ID _l U X u. < a vtru — l/> X x — »- 1- > o 101 © © o o C5 o o o o c o © © c © o o o o o o o o o o o o o o o o e o o o o o o o o o o o © o o c c o o o o o c o o o o © o n o o o o © o © © o o © o o © c o o o o o o o © o o a o © © o c o c o o o o o o o c o o o & Ll V LU f .— ~ © o © o o o o o mt «4 o O o o o 2 © o o o o © o o o o o o o o o V m o o in in in in o m in m in in m m cr (- O o o © © o o © o o © o o o o c. Z LL pri o o LO r- r- r>> o m in h- r» r- m r- u. I .* o o m -t * >*■ o in m *■ >»• ^ m >r J a. D r o ir u © o o o —* ■ri ~4 — < pd _ o o c o o c O o o o o o o o o o o o o © o m m m m m m m m m in IT IT in m m arf m «■! •* ^ m m m a f^ m in m in in m — o •o ci m o o © «m ri o o o o o o Z prf (\J o <0 lo iO ■0 m »l « <"1 « CI < f<"l < .<• « Z z z z z z z z z z Z Z z z z LU ffi UJ CO LL CD LU X > X > X > X > X > X > X > X > X > X > X >x > X >x > X OJ O LU C LU C LU C luO LU C LU C OJ O LU O ui O LU O UlO LU O LU O u. C JU -JO -1 o -JO _J o -J o -1 o _J o -J o -1 o _J o -1 o -J o -JO -J o 3 «/l a. 102 o o o o o e o o o o © o o © o o o o o o © o o o o o o o o o o o o o o o o o o o o o © o © o o e o o o o o o © o o © o © o o o o o o o o o o o © o © © o o o o o © © o o © © © © © © o o o o o o o o o © © o o © o Cl a. Z Ill DO o o o o o o o o o © o o o o o ac K o © o © o © o o o o o o o e o o Z (M © o o © o o o o o o o o © o o r UJ o © o o o o o © o o o o © o © u< z o o o o o o o o © c o o o c o D O c o e o o o IP CO 00 O O O m O o ir «o (M m o 0D CD O ao ec 00 IN o o If ir £ 00 fSJ ec « o o 00 CO O o IT o o oo 00 o o 00 00 (N IP o o ir\ 00 00 * (AN •o v4 pg (*> .# — UJ z o z o z c z o z o z o z o o z o z o z o z o z o z o z o I u < Ul UL 1— -*< z — « z z -*< z «VJ < z 1— fSJ < z *- z z (M ^ z (NJ < z ►- in < z 0- z z m < z z ID ec >x UJ O -JO UJ CC > X UJ O JU uj ec > z UJ C -I o UJ ec >z UJ C -JO UJ CD > z uj o -JO uj as > z Ul C -J (_> u. ec > Z UJ C JO Uj OC > z UJ C JU UJ ec > Z UJ O -JO uj er > z UJ C -JO Uj ec > z UJ c JO UJ ec > z uj o -JO uj cr > z UJ O -JO uj ec >Z uj O -JO Uj CC > X ui C -1 o 3 a t- 3 a 103 oooooooooooeoce ooooooooooooooo ooooooooooocoo o ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo n a z UJ tA Z m => o O o o — 4 mi *4 O mm ■4 o o o o o z o o o o o O O o O O o o o o o > U"> o o m tn in m o tn m m !/N in m in Of ►- m o o IA m in IT o IT m IP in in m m O 2 IM 0> o o o 0> 0» 0> ^ o> s> e» x UJ o o o 1* *4 mm «M o w im ** IM IM CM IM UJ X h» o o m 4 * —< mm mm mm mm mm mm z ^- IM IM fl r- K *~ IM m m f>. f- t^ m ^- o .* iO <0 •o * «»• ^ <0 in m ■* >r ** m •* mm • • • • • • • t • t t • • • • IB o o o o o O O o o o o o o o O UJ Of UJ -J UJ I o z o z c z o z c Z o z o z o z o z c z o z a z o z o z o z o < z •* < z z mm < Z IM < z nj< z IM < z IM« z IM < z IM * z fl < z z *> < z z 4- < z UJ CD UJ CD UJC > z > z > z UJ O uj O uj O -JO -JO -JO UJ CC Uj CD UJ Si UJ CD uj ar Ui 03 >z > Z >z > z >z > z uj o uj C UJ C UJ O UJ O uj o -iU -JO -JO _) o -JO -J o uj a* u- ar. uj or u. cd uj ac uj or >z >z >z >z >z >z uj C UJC uj O uj O UJ & uj O JO JO -JO JO -JO -JO 3 m- 3 O 10U o o o e o o o o o o o o o o o o o o o o o o o o o o o o o © o o o o c o o o o o o o o o o o o o o o o o o o o o © o o o c o o © o o o o o o o o o o © a o o o o o o o o o o o o o o o o o o o o o o o o o o o o o X o or 7 u a X mm 3 o o o © o o o o o o © c o o © Z o o o o o o o o o o o o o o © V © o o o o o o o o o o o o o o Of +- G o o o o o o o o o o © o o o o Z M O o o o o o o o o o o o o o o X U o o o o o o o o o o o o o o e UJ X © o o o o o o o o o o o o o c l/> * < ►- t- o e-< o o u. o o o o o o © mt — *4 © o o o o o o o o o o o o © o o o o o o o o o o o o o o o o o o o o o o * o> IM «m N IM <0 «M rg o> 0» 0> o ©> 0> o» c 0» o> z •0 ir> m IM •o <0 <0 * <#■ •* ■a >0 -o •4- ■c c >»■ m U"> - K ►- ►- f- k- t- »- >- ►- ►- «-« < -<< -* < — < IM < x >x > X > X > X >X > X > X > X > X >x > X > X > X > X uj r UJ O UJ O UJ C ui O uj O UJ O UJ O uj o uj O UJ C uj C UJ C UJ c Uj O JO -lij -JO -1 o -J o - o in O o o m o o m in o o in oo o o in o o in o o m o o in o o m o UJ X UJ z c z c z o CI z o I u u. o VI Z IM4 Z uj id uj oo >X >I uj o uj o -JO -JO UJ CO UJ CD UJ CO >z >x > x uj O uj O UJ O -jo jo -i o uj ec uj ec uj ec uj ec Uj CO ui ec UJ CO > z > z > z > z >x >x >x UJ O uj o uj o UJ O uj O UJ O uj o -JO -JO -JO -J o -JO JO -1 o UJ CD Uj CD UJ CD >X >X > X UJ O UJ C ui O -JO -JO -JO 3 io6 ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo X a o a> z r X 00 2 o o o o «* «-i o mt o o o o o o e »_ Z o o o o o e o O o o o o O o o in in in m ifi m m in in m m in in m m > ►- CM 80 (N m in h- m in o O iw o o o o o o o o z *. (Si 4 INI < (Si « (Si 4 (SI < IN* 1*1 < f»> < ZZZZZZZZZZZZ J •- J«- _i — J~ -J— J — J«- J"- -J >- -I >~ JW J«- U- CD UJ C UJffi UJCD UJOD UJ OC uj oc U- or UJ 0D UJ CD uj CD uj co >x >z >x >x >z >x >z >z >x >z >z >x uj o uj c in o uj o uj o uj o uj o uj o uj a uj o uio wo JU -JO -JO JO JU -JO -JO JO JO -JO -JO -JO ■* <. UJ CD UJ CD UJ CD >z >x >z uj o ujC uj O JU JO JO 3 D o 107 ooooooooooooooc ooooooooooooooo ooooooooooooooc ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo a. o X U-. X. z Z«M z UJ o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o g o o o o eooooooo ooeooooo ooeooooo oooooooo oooooooo oooooooo oooooooo o u. c z o LU ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo UJ r UJ z o z s z o z o X o u. o z z z z z z z UJ £C UJ UJ A > z >z >x UJ o uj O UJ o -JO JO JO UJCC UJ CO UJ CO UJ 0D uj er uj «e >X >x >r > x >z >x uu O UJ o uj o UJ O UJ O uj o JO JO JO J o JO JO Uj CD Uj CD UjCC UjCC U- CD U-jCD >z >z >r >z > x >z UJ O UJ O UJ O UJ O uj C Uj C JO JO JO JO JO JO < 3 a. ►- 3 o 108 ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo X tk o Ill z V* 3 o o o e o o o o o e e e o o o — Z o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o V K o o o o o o o o o o o o o o o ec z t}» t> 9> e> c* «> t> t> » «> 0* e* kCN <* IT in o» »• o e> 9~ e> » » o» a a o Z >z >z >Z > z >z > z > z > z >z > z >z UJ o UiO LU O uj c UJ O UJ O uj C uj a UJ o UJ o UJ O uj C -JO -JO JO JO -JO -1 o -I o JO JO JO JO JO uj ec u ec a cc > X > z > z uj C UJ O UJ O JO JO JO => a o 109 ooooooooooooooo ooooooooooooooo ooooooooooooooo OOOOOOOOOOOOOOO ooooooooooooooo ooooooooooooooo ooooooooooooooo JT UJ o CD z X => o o o o o o o o o o o o o o o «/> z o o o o o o o o o o o o o o o _ o o o o o o o o o o o o o o o ►- o o o o o o o o o o o o o o o >- Z c^ o o o o o o o o o o o o o o o ac LU o o o o o o o o o o o o o o o o X o o o o o o o o o o o o o o o X LU • • • • • • • • • t • • • • • UJ J o o o o o o o o o o o o o o o X UJ ac * o K h- < a UJ tx a u. o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o -J o o o o o o o o o o o o o o o pd o o o o o o o o o o o o o o o z o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o M • • t • • • • • • • • • • • t o o o o o o o o o o o o o o o o UJ or X LU J UJ X o -4 fVJ fO * w* 43 prf « «1 < n < tr\ < •r < z z z z z z z z z Z z z z Z z J — J — J«- J~ J~ J — J •- -J — J "- -J »- _j~* -J — -j •- -J — J «- LU CD uj ec LU 0C LU 00 UJ at lu ac lu an LU en Uj ec LU OB LU 00 U- CD LU 00 lu e LL 00 >z > X >z >z > X > X > r >x > X > X >x > X > X >x > X LU O LU O LU C LU C LU O LU O LU C uj O LU C LU C LU O LU C LU O LU O LU O -JO JO JU JO JU J o -J o JU -JO -J O -JO -JO J o JO J O t/1 3 < 3 a. 3 O 110 ooooooooooooooo oooooooocoooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo X in c tc z x 2 o o o o o o o o o o o o o o o (/» o o o o o o o o o o o o o o o — o o o o o o o o o o o o o o o k o o o o o o o o o o o o o o o V Z IM o o o o o o o o o o o o o o o a UJ o o o o o o o o o o o o o o o c X o o o o o o o o o o o o o o o r III • • • • • • • 1 • • • t • • t o *~ CL a. o z c o UJ z o o o o •o o o o a o o o o ©> o «SI o o o 0> o o o o o o o o V o o o o o o z o z o z a z a z o z o -r z o z u u. o UJ (E Uj It UJ fi UJ CC >x >z > x >x uj C uj O uj O UJ C JU JO JU JU uj (E ujc u. oc u cc > i >x > x > x LU C UJC uj O uj O -jo -jo ju ju uj cd uj cc uj a >x >x >x uj O UJ O uj O JO JO JO UJ CC Uj CO UJ flD UJCC >x >x >x >x uj C uj O UiO uj O jo jo jo jo 3 o Ill ooooooooooooooo ooooooooooooooo ooooooooooooooo ocooooooooooooo ooooooooooooooo ocooooooooooooo ocooooooooooooo o cc z x i o o e o o o o o o o e o e o e o o o o o o o o o e e o o o e f« o o o o o o e o o o e o o o o *- o o o o e o o o o e o o o o o > Z fM o o o o o o o o o o o o o o o of UJ o o o o o o o o o o o o o o o C X o o o o o o o o o o o o o o o X in • • t t t • • • • • • • • • • LU J o o o o o o o o o o o o o o o x u. QC • at & u. o o o o o o o o e o o o o o o o o o o o o e o o o m m IT *" m m in m m in m m m m m — •* • f» *• r o o» •* » o» o> e* o* o> o» z ■a ■o « &■ mm n. •m M p4 «* m* mm! m* wm —i o •* <*• # m •O >0 •O x > X > X >x > X >x > X > X >x >X > X > X > X > X > X UJ C ui o uj O ui O UI o UI O uj c UJ o UI O uj O uj O UJ o UJ O UI C uj O IL O JO JO JO JO JO JO JO JO JO JO J o JO JO J o J o 112 ooooooooeoooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo ooooooooooooooo X u. o SJ z T D o o o o o o o o o o o o o o o in Z o o o o o o o o o o o o o o o — o o o o o o o o o o o o o o o ►- o o o o o o o o o o o o o o o » Z«M o o o o o o o o o o o o o o o at Uj o o o o o o o o o o o o o o o o X o o o o o o o o o o o o o o o r UJ • • • • • • • • • • • • • • • o UJ or a. u. o o o o o tn m o o (* o o o o o o o o o o o m m o o ft o o o o o M"N in IT m m to «0 in in to in m m m in *• Ml art art rtj> art ■*■ h» art m-i ■* art art art art art — art 0> e* art 0> 9- (SJ c- 0« c- a »• z JT uj O -JO uj ec >x uj O -JO uj x uj O JO uj co >x ujc UJ <0 >x uj O JU z J — uj c UJ O J o uj cc u. flO UJ CC uj CO UJ CD > x >x >x >r >x UJ O UJC UJ C uj o uj o JO JO -JO JO JO UJ CD UJ C uj fi U- CD >x >x >x >x uj o uj C ujc UJC JO JO JO JO 3 Q. o 113 o e e e e e e e e e e 1 s e *- o e •* e • t e e o ' o o e « e e n © o « e> e e o o oo o o e $ z o O e — o o o o e e M MEG 0900 0000 0000 0000 ao 0000 » *• s e e e aft 1 e e e aft e © z ? ? * aft B.«o.*8 • • 1 9 i »■ -* « mm M • e I a> mi • >- in » ■O •»e o a o* e ■ e Be A i « UIM aft • in • o • ^g8§8-:.. -8 aft© o 2 Oat < ac o e e I o o o o © * Soeoo •efleS»x so o 5 o + o SO • o UI • r o h. b «. O OS UI e • O ►- © O u. o e go a e ■ o - .i •••• — • o a o z ION 00 00 00 00 REG OOt-lll O l- Uial «- ■ UJ K UI t UI ■— UI 1 M «ft •8 Z 3 u UI « m o >» ►- O o © o uj ac 3 — -I O I- •* * UJU * « » o o o at o Z OUJ *■ l/l > g z <•> «-?eo u. l»t»N © o o ■ uitr o o o C 3 Su UI UI 1 OK Cm j XcS ■M o i — o ■ -J O « >- 3 < o tm ►-.... U. — K u (• a> MUI UI O •M zo o e e x «e t- 2$ 1 m UJ uiac ac UJ o - 3 ac 3 18 u ao. U.U Si u 82 0C at >*■ »- 1- I- Z O UI*- ■ — at ■M MM ZUl o a at — — — — •-3 K u. »- K "• X z u. O It. 8 21II5S5 XO 52 ■ H Z *8 V u. ui ■ as as aa a> Z »/> >/» — e ee >. X 3 3 3 3 K 1- » ae UI UI UJ « ac <*at o ec >■ HHi»H53 ac « (/» ac ac a O ■ u. r UI s i U. 0k O Si ■J u UI 8" X UI 3 Z UI ae aeo Z UI uij J J jO« z Z U. X UJ Z ►- ui uj ui ui m » • < >- UI 3 > > > > _l -J s «/> H m O. UI UJ UI UI « a < i/l o 1- J _l _l _l z t- 3 O u. ■J _l D o M O o UI X O < at o a a. o a> « FormAEC-427 U.S. ATOMIC ENERGY COMMISSION vprw^ioi UNIVERSITY-TYPE CONTRACTOR'S RECOMMENDATION FOR DISPOSITION OF SCIENTIFIC AND TECHNICAL DOCUMENT ( Sm Instructions on Rmnrm Sid* I 1. AEC REPORT NO. C00-1U69-0231 2. TITLE FROG: DESIGN AND SIMULATION STUDY OF A MECHANISM WHICH LEARNS SELF PRESERVATIVE REACTIONS TO ITS ENVIRONMEN' 3. TYPE OF DOCUMENT (Check one): a. Scientific and technical report 1 I b. Conference paper not to be published in a journal: Title of conference Date of conference Exact location of conference _ Sponsoring organization □ c. Other (Specify) 4. RECOMMENDED ANNOUNCEMENT AND DISTRIBUTION (Check one): Kl a. AEC's normal announcement and distribution procedures may be followed. I | b. Make available only within AEC and to AEC contractors and other U.S. Government agencies and their contractors. | | c. Make no announcement or distrubution. 5. REASON FOR RECOMMENDED RESTRICTIONS: 6. SUBMITTED BY: NAME AND POSITION (Please print or type) Debasish Bose Research Assistant Organization Digital Computer Laboratory University of Illinois (Jrbana, Illinois 6l801 Signature 4/. Z7. 7%r77*c/€*£f-« Date August 6, 1973 FOR AEC USE ONLY 7. AEC CONTRACT ADMINISTRATOR'S COMMENTS, IF ANY, ON ABOVE ANNOUNCEMENT AND DISTRIBUTION RECOMMENDATION: 8. PATENT CLEARANCE: O a. AEC patent clearance has been granted by responsible AEC patent group. LJ b. Report has been sent to responsible AEC patent group for clearance. I~l c. Patent clearance not required. BIBLIOGRAPHIC DATA SHEET 1. Report No. UIUCDCS-R-73-588 4. Title and Subtitle FROG: DESIGN AND SIMULATION STUDY OF A MECHANISM WHICH LEARNS SELF PRESERVATIVE REACTIONS TO ITS ENVIRONMENT 3. Recipient's Accession No. 5. Report Date August, 1973 7. Author(s) Debasish Bose 8. Performing Organization Rept. No. 9. Performing Organization Name and Address Department of Computer Science University of Illinois Urbana, Illinois 6l801 10. Project/Task/Work Unit No. 11. Contract/Grant No. U6-26-15-301 12. Sponsoring Organization Name and Address Department of Computer Science University of Illinois Urbana, Illinois 6l801 13. Type of Report & Period Covered Thesis 14. 15. Supplementary Notes 16. Abstracts FROG is the simulation of mechanism which learns to discriminate between various classes of stimuli. The stimuli are some bugs (good and bad tasting) and predators. Each class of stimulus is characterized by the particular values of a set of visual characteristics. FROG acts like an adaptive filter. The visual data, interpreted through the filter, and a few somatic signals, which measures the requirements of FROG for feed and rest, are combined to decide upon one of the following list of actions: (l) feed, (2) rest, (3) flee. 17. Key Words and Document Analysis. 17a. Descriptors artificial intelligence cognitive system visual characteristics stimulus adaptive filter 17b. Identifiers /Open-Ended Terms 17c. COSATI Field/Group 18. Availability Statement FORM NTIS-35 (10-70) 19. Security Class (This Report) UNCLASSIFIED 20. Security Class (This Page UNCLASSIFIED 21. No. of Pages 118 22. Price USCOMM-DC 40329-P71 ocr 15 19? 3 \$l* ^ 510 84 Digital UNIVERSITY OF ILLINOIS-URBAN A IL6R no. C002 no. 588-589(1 973 computer Internal report/ 3 01 12 088400855 KHi IS J 111 H raw H 1 m QflSfinR 11 ■ M mm mm mm