THE LOGIC OF CAUSATION

THE LOGIC OF CAUSATION

Phase Two: Microanalysis

Chapter 11 –Piecemeal Microanalysis

1.Binary Coding and Unraveling

2.The Generic Determinations

3.Contraction and Expansion

4.Intersection, Nullification and Merger

5.Negation

1.Binary Coding and Unraveling.

We have developed a theory of causative propositions and arguments (eductions and syllogisms) by means of an analysis of the possibilities and impossibilities implied for the various combinations of the items concerned. This was characterized as ‘matricial analysis’, because of our recourse to tables for assessing and recording results.

But thus far we have only really engaged in elementary matricial analysis, which may be calledmacroanalysis. We shall now introduce a more advanced approach, which may be calledmicroanalysis. They are not different methods. Microanalysis is based on macroanalysis; it is merely a more detailed examination, digging deeper into the issues concerned, in an attempt to solve outstanding problems.

As we have seen, the determinations of causation are best expressed through amatrix, a table composed of ‘items’ and ‘moduses’. Theitemsare the terms or theses related by the causative proposition concerned. Each conceivable conjunction of these items, in positive or negative form, defines a row of the matrix. Themodusfor each such conjunction is a statement regarding its logical possibility or impossibility, or ‘openness’ (the latter in cases where the conjunction is in some unspecified contexts possible and in others impossible, so that an uncertainty remains). The moduses for the various conjunctions of items together constitute an additional column of the matrix.[1]

If we array the items of a matrix in a conventional arrangement (presenting the same row always in the same place), then the modus columns of all matrices will be comparable. By such standardization, we can express a determination of causation by merely writing down a string of moduses (i.e. its modus column), which we may call the modus of the determination concerned as a whole, or (for reasons we shall see presently) itssummary modus.

To simplify things, we may revert tobinary codes. We may express the presence or absence of each item in the matrix by a 1 or 0 notation. Similarly, we may code the modus for each row by a 1, 0 or∙(dot – meaning blank).The zeros or ones have different meanings in the items and modus cells of the matrix, note well:

Binary codes:

Such notation is merely convenient abbreviation, allowing us to express the summary modus of any determination as a relatively short string of digits and see the whole matrix in one sweep of the eyes. It is also, obviously, useful for computer programming purposes. Of course, if we are dealing with two items (say, P, R), the modus string will have 2²= 4 digits; if with three items (say, P, Q, R), it will have 2³= 8 digits; and so forth. Whether the string of digits isdistinctivefor each determination, we shall look into further on[2].

As we said, the rows of a matrix are defined and (conventionally) located by combinations of items. Thus, for two items, P and R, the four possiblePR sequencesare 11, 10, 01, 00, which may be labeled a, b, c, d if need be. We may choose this order of combinations as our standard arrangement (any other permutation is equally conceivable, but we conventionally settle on this one[3]). Similarly, for three items, P, Q and R, there are eight possiblePQR sequences, which may be labeled a-h if need be. And so forth, for more items.

We may thus, to begin with, present the matrices of the generic determinations of causation as in the following tables. These include (in the first two or three columns) the items in positive (1) or negative (0) forms, arrayed in standard combinations; followed by the summary modus for each propositional form (shaded column, symbolS),which you will recall we developed at the beginning of our research(in Phase 1, chapter 2)by analyzing the meaning of each of its constituent clauses and assessing the result of their interactions.

New columns are then introduced, which present all the conceivable realizations of the summary modus. These realizations, calledalternative moduses, are obtained simply by substituting, successively, a 0 (for ‘impossible’) or 1 (for ‘possible’) for each dot (‘open’ position) encountered in the summary modus, so that no dots are leftover. This process can be calledunraveling. The alternative moduses thus make explicit all cases inherent in the summary modus; and conversely, the latter is a summary of all the information contained in the former.

Note that the alternative moduses are themselves, ultimately, summaries, too. For while a zero (for impossibility) signifies that the combination of items concerned is in every context oralways absent, a one (for possibility) signifies that it is in some contexts orsometimes present[4]. Thus, to remove all implicit modality, and consider only actualities, we would have to dissect each such modus into an unspecifiable number of actualizations, where ‘0’ means absent and ‘1’ means present, simply. However, such further analysis is not needed for our purposes; the moduses as above defined are sufficiently informative.[5]

Consideration of a summary modus constitutesmacroanalysis; that of alternative moduses,microanalysis. That is all the difference between these two methods of matricial analysis: one of degree of detail. In the former, we have a rough idea of the relations involved; in the latter, it is as if we scrutinize them under a microscope.

The similar strings of zeros and ones used by computer programmers to code letters of the alphabet and symbols (I am thinking of ASCII codes) were arbitrary, pure conventions. But here, note well, once the meanings of zeros and ones, and the order of their presentation, are decided, there is nothing conventional about the string for each determination; it is alogicalproperty of it,objectively giveninformation.

2.The Generic Determinations.

In the four tables below, the precise significance of the numbers heading the columns of alternative moduses will be made clear in the next chapter; for now, just consider them as arbitrary labels. It should be stressed at the outset that thesemodus numbersare not to be confused with the determination numbers or mood numbers used in earlier chapters. Note also that Tables 11.1 and 11.2 concern two items (P, R), whereas Tables 11.3 and 11.4 concern three items (P, Q, R)[6]; the summary moduses of these two sets are therefore not directly comparable, the former being within a ‘two-item framework’, the latter within a ‘three-item framework’.

The two-item modus ofcompletecausation of form PR (symbolized bym,or more preciselym_PR) was previously established to be “10.1”. This is, through the following table, worked out to have two conceivable realizations, namely “1001” or “1011” (labeled respectively Nos. 10, 12).

Table 11.1.Matrix of “P is a complete cause of R”.

In contrast, the two-item modus, summarily put, ofnecessarycausation of form PR (symbolized byn,or most preciselyn_PR) was previously established to be “1.01”. This is, through the following table, worked out to have two conceivable realizations, namely “1001” or “1101” (labeled respectively Nos. 10, 14).

Table 11.2.Matrix of “P is a necessary cause of R”.

The three-item summary modus ofrelativepartialcausation of form P(Q)R (symbolized, according to context, byporp_rel, orp_Qor most preciselyp_PQR) was previously established to be “10.1.1..” . This is, through the following table, worked out to have sixteen conceivable realizations, as shown below (labeled respectively Nos. 149-152, 157-160, 181-184, 189-192). Note well that this is truerelativeto complement Q; we shall consider absolute partial causation further on.

Table 11.3.Matrix of “P (complemented by Q) is a partial cause of R”.

The three-item summary modus (columnS) ofcontingentcausation of form P(Q)R (symbolized, according to context, byqorq_rel, orq_Qor most preciselyq_PQR) was previously established to be “..1.1.01”. This is, through the following table, worked out to have sixteen conceivable realizations, as shown below (labeled respectively Nos. 42, 46, 58, 62, 106, 110, 122, 126, 170, 174, 186, 190, 234, 238, 250, 254). Note well that this is truerelativeto complement Q; we shall consider absolute contingent causation further on.

Table 11.4.Matrix of “P (complemented by Q) is a contingent cause of R”.

As above stated, we shall presently look into the summary moduses ofabsolutepartial and contingent causation. As we shall see, they are much wider than those relative to a given complement, dealt with above. This is natural, since absolute weak causations are vaguer forms than relative weak causations.

Comparing the summary moduses of complete and necessary causation with two identical items, 10.1 and 1.01, we see more clearly in what sense they are ‘mirror images’ of each other: the strings are identical, viewing one from left to right and the other from right to left. Similarly for the summary moduses of partial and contingent causation with three identical items, 10.1.1.. and ..1.1.01.

It should also be noted thata weak cause and its complement have the same summary modus. That is, partial causation of forms P(Q)R and Q(P)R have the same 10.1.1.. summary; and contingent causation of forms P(Q)R and Q(P)R have the same ..1.1.01 summary. This was obvious from the original definitions of these determinations, in which P and Q had the same relations to each other and to R; the distinction of one or the other of P, Q as complement was purely one of convenience or focus.[7]

Another observation we can make at this stage is thata generic determination and its (appropriate) converse would have one and the same summary modus.

That is true for the strong and absolute weak[8]determinations, which all concern two items. For instance, “P is a complete cause of R” and “R is a necessary cause of P” (note the change of determination, as well as that of item positions) are both here described by the string “10.1”. This can be ascertained by reading Table 11.2 in a different order, starting with the first row, then the third, then the second, then the fourth.

That is also true for the weak determinations, which involve three items. For instance, “P (complemented by Q) is a partial cause of R” is convertible to “R (complemented by notQ) is a contingent cause of P” (note well the change of complement polarity, as well as of determination and item positions) are both here described by the string “10.1.1..”. Again, we can prove this by rereading Table 11.4 in a different order, starting with the third row, then the seventh, then the first, then the fifth, then the fourth, then the eighth, then the second, then the sixth.

Indeed, we can say that convertibility is to be explained by such identity of moduses. Clearly, it follows that we cannot express direction of causation by reference to summary moduses. The orientations “from P to R” and “from R to P” must have some meaning – they are not empty verbal distinctions – but that meaning is not apparent in the way of a difference between moduses. It has to be sought in other properties, as already argued.

3.Contraction and Expansion.

Now, the above account does not allow us to compare the moduses of the strong determinations with those of the weak ones, nor tell us how to distinguish absolute from relative weak determinations. To enable such comparisons, we need to develop two processes: (a) contraction of a three-item modus into a two-item modus, and (b) expansion of a two-item modus into a three-item modus….

Let us first considercontractionof the three-item moduses ofporq(in their relative forms). Take first the case of partial causation by P of R, with reference to Table 11.3, above.

·The conjunction (P + Q + R) is possible, since the first row is always coded 1, whereas (P + notQ + R) is open, since the third row is sometimes coded 0 and sometimes 1. Nevertheless, it follows that the conjunction (P + R) is possible (i.e. to be coded 1), since “(P + Q + R) is possible” implies that “(P + R) is possible”. Note that if we regarded (P + R) as merely open, we would fail to record thatthere isnocolumn with 0s inboththe first and third cells.

Note this well:it is a finding we altogether missed in macroanalysis, and which may therefore affect some of our results.

·The same reasoning applies for the conjunctions (P + notR), comprising the second and fourth rows, and (notP + notR), comprising the sixth and eighth rows. They are both possible conjunctions, and not merely open.

·On the other hand, the conjunction (notP + R), comprising the fifth and seventh rows, must be declared open (i.e. be coded∙), since it is conceivable for both (notP + Q + R) and (notP + notQ + R) to be found impossible (as in the columns numbered 149, 150, 181, 182).

In this way the three-item modus for relative partial causation “10.1.1..” becomes the two-item modus “11.1”. Similarly with contingent causation: its three-item modus “..1.1.01” becomes the two-item modus “1.11”.

Thus, in case of need, we can contract a three-item modus into a two-item one, by changing a combination of 1 and 0, or 1 and∙, in corresponding locations, into a 1. Also, a combination of two dots yields one dot. Note well the rule of contraction:

1.Where there is a 1 in the 3-item modus, there must be a 1 in the 2-item modus.

Additionally note, though we have not yet encountered cases:

2.The only way we could obtain a 0 in a two-item modus, from a three-item modus, would be to findonly0s along both rows of the latter.

3.If we find cases of ‘11’,’10’ and/or ‘01’ mixed with cases of ‘00’ in the three-item modus, we must conclude a dot (∙) in the two-item modus.

Now, what have we found here? We started with weak causations by P of R, relative to some complement Q specifically, and ended with weak causations by P or R, without specification of Q, i.e.absolutely. The three-item modus forporqrelative to Q could not be equated to the same relative to some other complement, say Q₁; their matrices are superficially similar, but the items involved (namely PQR and PQ₁R) are quite different. But if we contract both kinds to two-item moduses, they would be indistinguishable, since the items involved (namely PR) are exactly identical.

Thus, the two-item modus ofabsolute(which includes relative)partialcausation of form PR (symbolized, according to context, byporp_abs, or most preciselyp_PR) is by contraction found to be “11.1”. This is, through the following table, worked out to have two conceivable realizations, namely “1101” or “1111” (labeled respectively Nos. 14, 16).

Table 11.5.Matrix of “P is a partial cause of R”.

In contrast, the two-item modus, summarily put, ofabsolute(which includes relative)contingentcausation of form PR (symbolized, according to context, byqorq_abs, or most preciselyq_PR) is by contraction found to be “1.11”. This is, through the following table, worked out to have two conceivable realizations, namely “1011” or “1111” (labeled respectively Nos. 12, 16).

Table 11.6.Matrix of “P is a contingent cause of R”.

Notice the similarity between the summary moduses ofm(10.1) andp_abs(11.1), or those ofn(1.01) andq_abs(1.11). Where for the strong determination we have a ‘0’ code, in the corresponding absolute weak determination we have a ‘1’; the remaining codes being identical. This, as we shall see in a later chapter[9], allows us to define the absolute weak determinations in formal terms.

Also note that the absolute weak determinations are convertible, just like the strong ones (as we pointed out in the previous section). For instance, “P is a partial cause of R” converts to “R is a contingent cause of P” (note the change in determination, as well as that of item positions), since these two forms have the same summary modus “11.1”.

The next question to ask is: what are the three-item moduses ofmorn, or ofporqin their absolute forms? We can answer this question by means ofexpansion, as follows.

Consider, to begin with, the strong determinations. In the case of complete causation by P of R, the following can be said:

·Knowing the conjunction (P + notR) is impossible, it follows thatboth(P + Q + notR) and (P + notQ + notR) are impossible conjunctions (whence the initial modus 0 becomes two moduses 0).

·Whereas, since the conjunction (P + R) is possible, it follows only thatat least oneof (P + Q + R) and (P + notQ + R) is a possible conjunction (i.e. they cannotbothbe impossible) – but we cannot predictwhichone is possible, so both conjunctions must be declared open (whence, the initial modus 1 becomes two moduses∙). Similarly,mutadis mutandis, for (notP + notR).

·Lastly, since the conjunction (notP + R) is open, so will a fortiori its two derivatives be (i.e. the initial modus∙becomes two moduses∙).

In this way the two-item modus “10.1” forcompletecausation becomes the three-item modus “.0.0….”. Similarly withnecessarycausation: its two-item modus “1.01” becomes the three-item modus “….0.0.”. Notice the loss of information occasioned by the change in each case, due to the fact that ones become dots; the results of such expansions are vaguer than their sources.

Thus, in case of need, we can expand a two-item modus into a three-item one, by changing a zero into two zeros in the appropriate locations, and a one or dot into two dots as appropriate. Here, note well the ‘appropriate locations’ arenotadjacent rows: they are the first and third, the second and fourth, the fifth and sixth, etc., reflecting a correspondence in combination of items – such as (P + notR) becoming (P + Q + notR) or (P + notQ + notR), which means moving from a PR sequence 10 to the PQR sequences 110 and 100, which signify the second and fourth rows of the matrix.

However, note well the restrictions implied in the followingrulesof expansion:

1. Where there is a 0 in the 2-item modus, theremustbe two 0s in the 3-item modus.
2. Where there is a 1 in the 2-item modus, therecannotbe two 0s in the 3-item modus.
3. Where there is a dot in the 2-item modus, there might be any combinations of 0s and/or 1s in the 3-item modus.

With regard to ‘zero’ moduses (impossibility), they are universalized as it were from the initial row to the corresponding expanded rows. With regard to ‘ones’ (possibility), what is universalized from the single initial row to the two subsumed rows is the interdiction of zeros: just as in the two-item modus 0 is excluded by 1, so the three-item expansion cannot include columns (moduses) having 0s in both the corresponding rows. A fortiori, in the case of ‘dots’ (which might include zeros or ones), we cannot predict combinations in the two cells concerned, since all pairs are allowed, i.e. 0 and 0, 0 and 1, 1 and 0, 1 and 1.

Consider now the weak determinations, in accord with the rules of expansion just ascertained. If we similarly expand the two-item modus of absolute (or relative)partialcausation, namely 11.1, into a three-item modus, we obtain “……..”, since all ones or zeros become dots. Likewise, by expansion of the two-item modus of absolute (or relative)contingentcausation, namely 1.11, into a three-item modus, we obtain “……..”.

Note well that the result in both these cases is a string of dots, signifying complete uncertainty, the least possible amount of information. Each initial one or dot was expanded into two dots, so that all remaining specificity in the initial string was dissolved in its derivatives.

Note also the marked difference between the three-item strings of the absolute weak determinations “……..”, and those of the corresponding relative forms, namely “10.1.1..” and “..1.1.01” respectively.

Clearly, for all four generic determinations, expansion of a two-item modus “1” (possible) into two three-item moduses “∙“ (open) results in aloss of data; i.e. the information that ‘at least one of the two conjunctions concerned must be possible: i.e. they cannot both be impossible’ is no longer coded in our table. A calculus of causation should be so designed as to avoid all loss of information due to mere linguistic inadequacies[10]. Thus, we have to find a way to express, through a special code in the modus, say µ (Gk. letter mu), thatat least one of the two (or more) conjunctions so coded is implicitly a “1”.[11]

This measure by itself is not enough; to save all available information, we would have to specify the rows concerned, say by labeling them a-h. For instance, if at least one of rows ‘a’ and ‘c’ has to have modus 1, each would have to be coded more specifically as µ_ac. Such coding means that the PQR sequences signified by the labels a and c (namely, 111 and 101) may have moduses with 0 and 1, 1 and 0, 1 and 1 – but they may not have the pair 0 and 0. It follows that:

·if a = µ_acand c = 0, then a = 1 (since 00 is inconceivable), and

·if a = µ_acand c = 1, then a =∙(since 01 and 11 are both conceivable);

and likewise, of course, if c = µ_acand a = 0 then c = 1, and if c = µ_acand a = 1 then c =∙.

All this information can be considered implicit in a table like the following (the relative weak determinations of form PQR are included for comparison):

Table 11.7.Summary moduses for the six generic determinations of form PR or PQR.

All this may seem pretty complicated, but as we shall see it simplifies a lot of things. Through the summary moduses in the above table, we can identify precisely the alternative moduses in a three-item framework implied by each of the four determinations (for the items PR or PQR).

As we shall see in the next chapter, the strong determinationsmandnturn out to have 36 such alternative moduses each, while the weak determinationspandqin absolute form (as here), have 108 alternative moduses each (to compare to the 16 moduses of relative weaks). We shall list these moduses in the next chapter, so no need to do so here; they are easy to unravel by substituting zeros and ones for dots as previously explained.

4.Intersection, Nullification and Merger.

We shall now consider certain inferences from the above data.

The joining of generic determinations can be considered as theintersectionof their respective summary moduses. By such conjunction of two propositions (or more), two classes (generic determinations) are used to express a more restrictive class (a joint determination), with whatever they havein common.

By this process, in atwo-itemframework (where the weak determinations are absolute), the joint determinations are found to have the following summary moduses:

·complete-necessary causation,mn= 10.1 + 1.01 = 1001 (modus No. 10);

·complete-contingent causation,mq= 10.1 + 1.11 = 1011 (modus No. 12);

·necessary-partial causation,np= 1.01 + 11.1 = 1101 (modus No. 14);

·partial-contingent causation,pq= 11.1 + 1.11 = 1111 (modus No. 16).

As can be seen, the result in each case is a single alternative modus (mentioned in brackets), which represents what the joined generics have in common. Thus, for instance,mhas moduses 10 and 12, andnhas moduses 10 and 14; thereforemn(meaningm+n) will have modus 10. The resulting summary modus is more defined than its sources, i.e. there are less dots, there are less uncertainties in the relation between the items.

This operation is merely an application of the well-known rule of class logic, that the logical product of two classes (such asmandn, each of which subsumes two subclasses, namely 10 and 12 formand 10 and 14 forn) is the elements they have in common (namely, modus 10, in the case ofmn). This can be seen for example in an Euler diagram, comprising two circles which overlap: theircommonarea is the outcome of their product, and usually smaller than the circles (in our example, modus 10).

Note that by ‘logical product’ logicians mean that the two (or more) classes are conjoined together (i.e.mnmeansm + n)[12]. It must be stressed that modus lists are disjunctive not conjunctive, so that underlying this formula is another one (namelymn= ‘modus 10 or modus 12’ and ‘modus 10 or modus 14’, which means ‘in any event, modus 10’, i.e. it refers to the leftover after removing from consideration the elements ‘modus 12’ and ‘modus 14’, which are exclusive in either disjunct.

Similarly, in athree-itemframework (where the weak determinations may be absolute or relative), intersection of the generic determinations yields the joint determinations, with the following summary moduses:

·complete-necessary causation:

·complete-contingent causation,

·necessary-partial causation:

·partial-contingent causation:

Here again, the result signifies the alternative moduses that the joined generics have in common; we shall not list them at this stage: the list will be given in the next chapter. In the case ofp_relq_rel, exceptionally, the result is a fully specifying summary modus, i.e. a single alternative modus (that labeled #190, as we shall see later). The resulting summary modus fuses together the most definite elements of the initial summary moduses; some dots become µ’s, and some dots or µ’s become more specifically a 0 or a 1. The µ’s concerned are in pairs like µ_acremember; the subscripts are not mentioned here for brevity.

Some of these results correspond to those obtained by macroanalysis, note. To grasp therulesof intersection, let us review the examples shown above:

1. The summary moduses are never conflicting in a given position (1 in one case and 0 in the other); this simply means that the determinations joined arecompatible.
2. For each position identical in both generic summary moduses, or more definite (µ or 1 or 0) in one and indefinite (∙or µ) in the other, the resulting corresponding position in the joint summary modus has that equal or more definite value.

Wecannotjoin two determinations whose summary moduses have conflicting elements in the same position (a 0 in one and a 1 in the other): they are incompatible propositions, it is an impossible conjunction. In alternative modus terms, it means that these determinations do not have even one modus in common; in class logic terms, it means that the given classes (generic determinations) do not overlap: they have no intersection. Such logically empty concepts are known as null classes; we might therefore refer to the act of judging a class to be null asnullification.

For instances, the conjunctionsmp,nqare null classes. Sincemhas moduses 10, 12 andphas moduses 14, 16, they have no common ground, no modus in which to coexist. Similarly fornandq,mutadis mutandis. This we know already from macroanalysis. More interesting, isthe capacity nullification gives us to judge the feasibility of lone determinations, as we shall see in the next chapter.

Let us now consider another logical composition, that ofmerger, which disjoins two (or more) propositions, to obtain a single, vaguer proposition. In alternative modus terms, this process puts together all the alternative moduses listed for the given propositions in a larger list for the merged proposition. In class logic terms, this means that the two (or more) classes together become a single class covering all the areas they have exclusively as well as those they have in common.

This corresponds to the ‘logical sum’ of classes, where the two initial classes merge into a larger class by inclusive disjunction (expressed by operatoror, which here means ‘and/or’, i.e. ‘not both not’; this is often symbolized by a ‘v’, or in some computer languages by a ‘½‘)[13]. Inclusive disjunction means that all the elements subsumed by the given classes are to be included in the larger class; if a subclass subsumesxelements and another involvesyelements, then the larger class covers (x + y) elements. In contrast, in conjunction, only the elements subsumed by all the given classes are selected, forming a narrower class.[14]

We can, for instance, merge joint determinations into generics; thus, “mn v mq” is equivalent to just “m”, “mn v np” becomes “n”, “np v pq” becomes “p”, and “mq v pq” results in “q”. We can likewise merge generic determinations into broader concepts, such as strong or weak causation or causation, as shown below. Merger is easy if we work directly with alternative moduses; but it becomes very complicated if we refer to summary moduses, due to the inadequacies of thead hocnotation system we have used so far.

In atwo-itemframework, it is feasible if we introduce an additional symbol, sayl(Gk. letter lambda), signifying that the two positions in the formula where it occurs cannot both be coded ‘1’ (in contrast to µ, which signifies that they cannot both be ‘0’, remember). In such case, we can predict the summary moduses of the following vague propositions (s,w,c) on the basis of the generics merged in them:

·strong causation, symbols=m or n= 10.1 v 1.01 = 1ll1 (moduses Nos. 10, 12, 14)[15];

·absoluteweak causation, symbolw_abs=p_absor q_abs= 11.1 v 1.11 = 1µµ1 (moduses Nos. 12, 14, 16);

·relativeweak causation, symbolw_rel=p_relor q_rel= sametwo-itemsummary modus as for absolute weak causation;

·causation, symbolc=m or n or p_absor q_abs= 10.1 v 1.01 v 11.1 v 1.11 = 1..1 (moduses Nos. 10, 12, 14, 16).

Note that ‘causation’ here meanssomecausation, causation ofanydetermination whatever, whetherm,n,p_absorq_abs. As we will show in the next chapter, ‘contributory causation’ (m or p) and ‘possible causation’ (n or q) are different from it only with reference to relatives; in absolute terms, they are identical to each other and to causation (becausemimpliesnot-p_abs, andnimpliesnot-q_abs).

The same four operations in athree-itemsystem all apparently yield one and the same conclusion, namely “µ.µ..µ.µ” (try and see) – which is the summary modus of causation, covering 144 alternative moduses, as we shall see. This is of course an absurd result, because, as we shall see in the next chapter, strong causation in fact covers 63 moduses; absolute weak causation, 135 moduses; and relative weak causation, 31 moduses! It follows that our notation system is inadequate for merger operations other than:

c= µ0µ0.µ.µ v µ.µ.0µ0µ v µµµµ.µ.µ v µ.µ.µµµµ = µ.µ..µ.µ (144 moduses).

What this means is that the symbolic language developed so far is too simple to express more complex relations than those intended by a 0, 1,∙or µ (or evenl, just introduced to enable merger of the two-item summary moduses of strong determinations[16]). It does not generate a distinctive summary modus for each and every form. However, I will not bother to attempt improving on it, not wishing to get bogged down in inessential matters. For our primary goal here is not to develop a calculus of summary moduses, but to ascertain how generic propositions can be merged into vaguer forms. And this we can readily do with reference to the underlying alternative moduses, which is good enough.

5.Negation.

Before moving on, let us review the ground covered thus far. We started withbinary codingof the summary moduses of the generic determinations known to us thanks to macroanalyses performed at the very start of our research into causative propositions. We saw that these summary moduses involved uncertainties (coded∙). To eliminate these information gaps, we had tounravelthe summary moduses, that is, identify the underlying alternative moduses (involving 0 or 1 codes exclusively). We thus introduced microanalysis.

However, the strong determinationsm,nwere expressed in a two-item (PR) framework, while the relative weaksp_rel,q_relwere expressed in a three-item (PQR) framework – so these two sets of forms were not comparable. We therefore had to work out the means forcontractionandexpansionof their summary moduses (the latter process required that we introduce a fourth code, µ). This also allowed us to ascertain the two- and three- item summary moduses of absolute weak determinationsp_abs,q_abs– first by contracting those of the relative weaksp_rel,q_rel, then by expanding these results.

Having thus obtained both the two- and three- item summary moduses of all six generic determinations, we had all the information we need to work out the matrices of all derivative propositions. Indeed, by means ofintersectionwe can readily identify the alternative moduses of any conjunction of determinations: they are the alternative moduses the latter have in common. A special case of this isnullification: if the propositions we wish to conjoin have no alternative moduses in common, they are incompatible. And by means ofmergerwe can readily identify the alternative moduses of any disjunction of determinations: they are the alternative moduses the latter have all taken together.

We thus dispose of the basic data and logical processes we need for microanalysis of all positive forms, be they generic, joint (i.e. narrower than the generics) or vague (i.e. broader than the generics). But we still lack the alternative moduses of negative forms of whatever breadth. We cannot obtain their summary moduses by macroanalysis, as we did for the generic positive forms, because of the underlying complexity of negative causative propositions. So we must look for more profound means.

Thus far, we have engaged in microanalysis that may be characterized aspiecemeal. In the next chapter, we shall approach this topic with a more holistic perspective, which we may refer to assystematicmicroanalysis. That consists in considering all conceivable alternative moduses in a given framework (fixed by the number of items under consideration), and then locating the determination(s) under consideration within this full range of possibilities.

The alternative moduses of negative forms become easy to identify thereby. Having the list of all conceivable alternative moduses in a given framework, and the alternative moduses of a positive form, we can readily infer those of the corresponding negative form: they are all the remaining alternative moduses! This process, which we shall simply callnegation[17], is akin to subtraction. If a class subsumesxelements and a subclass of it involvesyelements, then the remaining area covers (x – y) elements.

Microanalysis thus ultimately enables us to distinctively define any and every causative proposition (and other, related forms, as we shall see), with little effort. Furthermore, such detailed matricial analysis turns out to be a panacea, providing us with resolutions to all deductive issues in causation.

In particular, note that once we identify the moduses of negative generics, we can ascertain those oflonedeterminations, which conjoin one positive generic with the negations of all others. As we shall see in the next chapter,absolutelones are nullified. However, as we shall see in a subsequent chapter,relativelones are not nullified. Let us here mention for the record theirsummarymoduses, which may be constructed knowing their alternative moduses, there identified (check and see for yourself that these summaries give rise to the correct alternatives):

·m-alone_rel= µ0µ0µµµµ

·n-alone_rel= µµµµ0µ0µ

·p-alone_rel= 10.1µ1µ.

·q-alone_rel= .µ1µ1.01

If we compare these to the summary moduses ofm,n,p_relandq_rel, respectively (which are given in Table 11.7 above), as well as to those of joint determinationsmn,mq_rel,np_rel,p_relq_rel(given in the previous section), we may observe the following mutations.

A code 0 or 1 for a generic is retained in a joint or lone including it. Aµfound inm(orn, as the case may be) is retained inmn, and inm-alone_rel(orn-alone_rel), but not inmq_rel(ornp_rel), because in the latter oneµis superseded by the 1 found in the corresponding position inp_rel(orq_rel), so that the remainingµbecomes a dot. There are no dots left inp_relq_relbecause all the dots inp_relorq_relhave all been superseded by a 1 or 0. A dot inmornbecomes aµinm-alone_relorn-alone_rel, respectively. As forp-alone_relorq-alone_rel, the dots inp_relorq_relnotpaired-off with a 1 becomeµ, whereas those paired-off with a 1 remain dots.

As already explained, aµsignifies that the pair of cells containing it (the first and third, the second and fourth, the fifth and seventh, or the sixth and eighth) may separately be 0 or 1, but cannot together be 0. No such restriction occurs where there are mere dots. Thus, what the above teaches us, especially, is that a relative lone determination has aslightly more restrictivemodus than the corresponding generic determination, but is in all other respects identical.