The Machine Scientists: Iatrophysics and Selective Scientific Realism

The Machine Scientists: Iatrophysics and Selective Scientific Realism

The Pessimistic Meta-Induction on the History of Science (henceforth ‘PMI’), as developed by Laudan (1981)1, challenges the consistency of Scientific Realism with the history of science. I examine Stathis Psillos’ (1999) form of moderate scientific realism, which attempts to avoid the difficulties of PMI.

Part I examines the PMI and a reformulation as a ‘Pessimistic Reductio ad Absurdum’ (henceforth ‘PRA’). A successful defence against the PMI-formulation is offered, in light of which the PRA-formulation is emphasised. I then expound Psillos’ realism, based on his Divide et Impera stratagem, as a realist position responding to the challenges of the PRA.

In Part II, I present a case-study from the history of medicine, specifically the 17th-18th century physiologico-medical Iatrophysics movement. This case provides a typical counterexample to the realist theses via PRA. Following Psillos’ strategy, I present a moderate realist account of this historical case, demonstrating that Iatrophysics is no counterexample to sophisticated realist positions, such as Psillos’.

Part III defends Psillos’ position against three objections: that (i) Psillos’ position is ad-hoc, (ii) Divide et Impera is incompatible with realism, and (iii) we cannot decide which constituent claims of a theory we should believe are approximately true.

I: The PMI and Selective Scientific Realism

We first require a putative definition of the Scientific Realist’s position. Van Fraassen (1998, p.1065) characterises naïve realism by:

SR1:    Successful scientific theories are true descriptions of the world.

SR1 is clearly unsatisfactory (Lyons, 2009, pp.66-7); history abounds with successful scientific theories which are no longer deemed true – Newtonian mechanics, for instance. Rather, we take a weaker claim:

SR2:    Successful scientific theories are approximately-true (or ‘truthlike’) descriptions of the world. (Psillos, 1996, p.S307)

 PMI attacks SR2 by undermining the link between scientific success and truthlikeness. Laudan proceeds through his ‘Historical Gambit’; he presents a catalogue of past scientific theories which, despite their successes, were since rejected in favour of incompatible theories (1981, p.33). Laudan’s list includes humoral medicine, Ptolemaic astronomy, optical aether, Noachian geology and vital-force physiology. The list is not exhaustive – Laudan believes it could be “extended ad nauseum” (ibid.). Thus, he inductively infers that even our most successful theories may be superseded by incompatible alternatives, despite their successes. We conclude that success cannot guarantee truthlikeness for scientific theories.

An effective response is to challenge Laudan’s attributions of ‘successfulness’. Laudan’s ‘success’ threshold is too easily reached, unduly inflating his inductive evidence. First, he understands success as explanatory successes (1981, p.30). Saatsi (2009, p.360) objects to this commonplace conflation; scientific realists should instead hold that predictive success is crucial in defining successfulness for SR2. A theory may be able to explain the phenomena in its domain, but this alone does not qualify that theory as successful for the realist’s definition. For instance, Intelligent Design can explain any phenomena as the idiosyncratic creation of the Designer, but is not therefore scientifically successful. As Laudan’s criterion of successfulness is explanatory success, his catalogue of successful yet false theories is not entirely relevant to SR2. Rather, we need a revised set of predictively successful theories. The new list is shorter – humoral medicine and Noachian geology, for instance, may arguably be excluded.

Furthermore, the realist is wisely sceptical of predictively successful theories which only predict results engineered into the theory. One can contrive theories to entail known phenomena. A sophisticated Intelligent Design theorist could specify the Designer’s penchants precisely to predict the observed phenomena. To qualify as predictively successful, the results predicted must be somehow ‘novel’. We cannot hold, however, that only temporally novel results count as predictive successes, as this unpalatably entails that a theory’s successfulness is dependent upon the time at which it was proposed. Rather, following Worrall (2002, pp.194-6), we can adopt a use-novelty definition of predictive success. Evidence E is a use-novel predictive success of theory T if and only if T predicts E, E is empirically corroborated, and E was not used in the construction of T. Thus, E need not be discovered temporally later than T to be a predictive success of T; the fact that T predicts E, without being deliberately engineered to do so, suffices. A case which Worrall envisions as exemplifying use-novel predictive success is Einsteinian Relativity’s ‘prediction’ of Mercury’s perihelion precession; the empirical fact was known prior to Einstein stating his theory, but the result naturally follows from the theory, without adaptation to yield it (see Redhead, 2001, p.342).

Finally, scientific realists commit to SR2 only in relation to mature scientific theories. This move may seem ad hoc, defining away the issue by rejecting as immature those theories threatening SR2. However, Worrall (1996) provides independently testable criteria of maturity to defuse this objection. A theory is mature if and only if it makes use-novel predictions, and these are empirically corroborated (pp.153-4). This parallels Lakatos’ (1968, p.164) notion of “progressive” Research Programmes. For Lakatos, a programme is progressive if and only if it is both theoretically and empirically progressive. Programmes are analysed with respect to the ‘problemshifts’ they undergo; these occur when empirical evidence challenges the theory, and modifications are made. Theoretically progressive programmes have, at every problemshift, excess predictive content over their predecessor. Empirically progressive programmes have their excess predictive content empirically corroborated (ibid.). This definition of maturity allows the scientific realist to screen-out ‘degenerative’ programmes and theories, which make only ad hoc modifications to evade counterevidence, from their set of successful theories. So, we revise SR2 to:

SR3:    Mature scientific theories are approximately true.

After these defences, Laudan’s list is radically diminished. However, Worrall notes that we cannot hope that the set of mature successful theories now regarded as false will be empty, taking Fresnel’s wave optics as an example (Worrall, 1996, pp.154-5). Psillos additionally accepts the caloric theory of heat as a rejected mature successful theory (Psillos, 1999, p.108). I will argue (Section II.2) that the Iatrophysical physiology is also a contender to fulfil these conditions.

An inductive argument like PMI is weakened if its inductive base is small. But PMI can be reformulated as a Pessimistic Reductio ad Absurdum (PRA) of SR3, as Psillos acknowledges (1999, pp.102-3). This formulation threatens SR3 with any single counterexample. The argument proceeds:


Suppose: SR3.

1) Suppose T1 and T2 are mature successful scientific theories of similar domain, such that T2 supersedes T1, and T2 denies the existence of an entity e which is postulated by T1.
2) T1 and T2 are both approximately true.                     [by SR3 and (1)]
3) The claim “e does not exist” is at least approximately true.     [by (2)]


4) The claim “e exists” is radically false (i.e., not even approximately true).

T1 entails “e exists” by (1), hence:

5) T1 is radically false.

There is a contradiction between (2) and (5), so by Reductio ad Absurdum, SR3 falls. Unlike PMI, all the anti-realist requires for this argument is a single pair of theories, T1 and T2, satisfying the criteria for (1). This would provide a refutation to SR3. Possible pairs are clear from Laudan’s list. For instance, Fresnel’s wave optics (T1) postulated a luminiferous ether (e), which was rejected by the superseding theory of 19th-century electromagnetic optics (T2) (Worrall, 1996).

Psillos’ moderate realist position is developed in response to the PRA, reformulating the realist thesis to expose the weaknesses in the argument. First, the inference from (2) to (3); Psillos’ rejects this move, arguing that by claiming ‘T is approximately true’, we do not ascribe approximate truth equally to all “Constituent Claims” (Psillos, 1999, p.108) of T. Rather, by claiming the approximate truth of T, we claim that at least some subset of the constituent claims of T are truth-like – in Psillos’ system, those most strongly confirmed by T’s predictive successes.

Second, the inference from (3) to (4); while it may be the case that e does not exist, intuitively the claim “e exists” might be approximately true if some f which is very similar to e exists. The claim that e exists is strictly speaking false, yet approximates the true situation. For instance, Psillos (1999, pp.137-40) argues that the postulate of a luminiferous ether in Fresnel’s wave optics was sufficiently similar to the electromagnetic field that, if the postulate of the electromagnetic field is true, the postulate of a luminiferous ether is approximately true.

Thirdly, the inference from (4) to (5); again, Psillos’ position allows T to have false consequences, and yet possibly remain approximately true. Newtonian mechanics, for example, has false consequences in predicting Mercury’s perihelion precession, and yet is still regarded as a good approximation to the truth (Einstein, 1916, pp.193-200).

To actualise these arguments, we require a fine-grained realist position which differentiates between a theory’s individual constituent claims. Psillos presents his strategy Divide et Impera to provide this.

Psillos draws on Kitcher’s (1993, p.149) similar position; Kitcher rejects the assumption, which he calls ‘Blockish Holism’ (Kitcher, 2001, p.170), that when T is rejected, all of its constituents are rejected too. If T is false, it does not follow that any given constituent of T is false too. Given this, we see that when T2 supersedes T1, there may be a subset (call it T1*) of the constituent claims of T1 which are consistent with T2. Furthermore, a subset of T1* may be retained in T2. Divide et Impera consists in identifying the compatible subset T1* of T1 (Psillos, 1999, p.108).2

Psillos divides the constituents of T1 again, this time isolating the subset of T1 (call it T1′) responsible for the predictive successes of T1. Psillos notes that not every constituent of a theory is responsible for a given prediction. We might call T1′ the ‘successful constituents’ of T1. Sophisticated realists need only commit to the successfulness of T1′, and hence only to the truthlikeness of T1′. The crux of the argument is that a predictive success supports each consistuent claim of a theory to a different extent, depending on how integral the claim was to the prediction (ibid., pp.109-111). We should have ‘differentiated belief’ (Psillos, 1994, p.161) in constituent claims based on the evidence for each claim. Suppose I have a bag of objects, and a theory T with the constituent claims: (1) ‘If x is a red object in the bag, x is a sphere’, and, (2) ‘If x is a blue object in the bag, x is a cube’. Naturally, drawing a red sphere from the bag supports T, and confirms both claims 3, but confirms (1) to a greater extent. If we repeatedly draw red spheres, (1) will be confirmed far more than (2), and we should believe (1) more strongly than (2).

An obvious objection, that the other claims are nonetheless involved in the derivation (Worrall, 1994, p.337), is not pertinent to Psillos’ argument. Unlike Kitcher’s (1993, p.149) position, which distinguishes “working” from “idle” postulates, Psillos does not deny that theoretical claims which are later rejected play a role in the theory’s successes. Rather, he denies that they take a crucial role, “fuelling” the successes (Psillos, 1999, p.110). T1′ is the set of constituents which are most strongly supported by the evidence for T1.4

We now reformulate our realist thesis once again:

SR4:    Mature successful scientific theories have some truth-like constituent claims. (Psillos, 1999, p.109)5

The anti-realist may argue that the retention of some constituents of T1 in T2, and the observation that some claims (T1′) of T1 are more supported by the evidence than others are merely historical artefacts. However, Psillos now makes a strong Historical Claim, HC:

HC:     T1′ ⊆ T1*

In words, the successful constituents of the superseded theory are a subset of (or equal to) the set of constituents compatible with the superseding theory (Psillos, 1999, p.110). If HC holds, then those claims which the realist deems truth-like by SR4 are consistent with the change to T2. The claims SR4 holds are truth-like are not those which T2 contradicts, and hence there will be no entity e postulated by a constituent of T1′ which is denied by T2.

Let us re-evaluate PRA. We now hold SR4, not SR3. So, the successfulness of T1 and T2 (postulated in (1)), allows us to infer in that there are truth-like constituents of T1 and T2. First, this cannot license the initial step of the argument, that ‘e does not exist’ is at least approximately true. We would need further evidence that the claim ‘e does not exist’ is one of the truth-like claims of T2 (i.e., a member of T2′) in justifying that step in the argument.

But suppose this is the case. Even so, this would only licence the claim that T1 taken as a whole is radically false. But we have seen the error of ‘Blockish Holism’. Unless T1′ either contains the claim ‘e exists’ or entails that claim, the argument does not compel the conclusion that T1′ contains a radically false claim.

In other words, Psillos’ Divide et Impera move requires the PRA proponent to establish two further premises:

  1. e does not exist’ is a successful constituent of T2 (i.e., a member of T2′)
  2. e exists’ is part of, or implied by, T1′.

Per Psillos’ Historical Claim, it will not be the case that both of these premises hold true together. Thus, if HC is true, the PRA is blocked.

However, HC is an empirical claim about the history of science, and is far from obvious. We can only uphold Psillos’ defence against the PRA if the historical record supports HC for each putative T1 in the set of rejected mature successful theories. Psillos (1994; 1999) attempts to show this for the caloric and optical ether cases. It is not sufficient, though, to focus solely on the clear-cut examples from the physical sciences. HC ranges across all fields of science, and has not been tested against, for example, medical and physiological cases. Laudan’s catalogue of rejected successful theories includes several from these fields, including the “vital force theories of physiology” (Laudan, 1981, p.33). In section II, I will assess Psillos’ realism, and HC, with reference to a similar (though more successful and thus more challenging to scientific realists) physiological theory – Iatrophysics.

II: Iatrophysics

            The soundness of HC is critical to Psillos’ attempt to defend SR4 from the PRA. HC is an empirical hypothesis, requiring justification in those cases forming Laudan’s catalogue; that is, any pair of theories T1 and T2, such that T1′ contains some constituent claim x which is incompatible with T2 (i.e. x is in T1*). One potential case is Iatrophysical physiology, which flourished in the 17th-18th century, following Descartes’ Treatise of Man (1664). To assess whether HC is substantiated in respect to Iatrophysics, we must present the programme, T1, (Section II.1), demonstrate that it was successful in the sense intended by scientific realists (II.2), and identify the constituent claims responsible for the predictive successes, T1′ (II.3). Finally (II.4), I analyse whether T1′ is a subset of the set of claims, T1*, compatible with modern physiology (here taken as T2).

II.1: The Iatrophysics Programme – T1

Descartes’ Treatise of Man (1664) is considered the foundation of Iatrophysics. The inspiration for the programme came from the contemporaneous successes of Galileo and Kepler, amongst others, in physics. The attempted axiomatisation of physical phenomena posed a challenge, to Descartes, to align biological science with the mechanistic analyses favoured in physics (Singer, 1928, p.127).

Iatrophysics holds that bodily phenomena can be analysed mechanically, through applied physics and mathematics. Iatrophysicists shared certain methodological principles – that each organ is mechanical, and so the organism can be analysed as a complex machine (Shryock, 1961, p.221), and commitment to corpuscularism; all processes should be explained by the movement of material particles or corpuscles (Hall, 1972a, p.xxvii).

Descartes provided a mechanical theory of locomotion, which initiated Iatrophysical physiology. He advocated the ‘Heart-Heat’ theory, a long-held doctrine that the heart, by fermentative processes in the left ventricle, heats the blood (1664, p.9-15). This is consistent with the mechanistic precepts of Iatrophysics, as the fermentation is a motion of blood corpuscles (Hall, 1972b, p.xxxvi). Using hydraulically animated statues as inspiration (1664, p.22), Descartes develops a pseudo-hydraulic theory of locomotion (Starobinksi, 1964, p.51). The finest blood corpuscles, named ‘Animal Spirits’, rise to the brain (1664, p.17). Nerves are hollow tubes encasing a thin string, linked to brain cavities, and filled with Animal Spirits (p.24). When the nerve receives stimuli, the string pulls on the brain matter, causing an opening in the brain at the top of the nerve to widen, and an influx of Spirits (Pubols, 1959, p.114). As the nerve-tubes are full, this causes an immediate displacement of spirits into the muscle at the other end. This influx of new material into the muscle causes contraction, explaining the extra bulkiness of contracted muscle (Descartes, 1664, p.21). Involuntary motion is thus explained through a primitive reflex arc. Voluntary motion is caused similarly; the Pineal Gland, which Descartes hypothesised was the seat of the soul, influences the flow of spirits into certain nerves to cause locomotion (Brown, 1971, p.63). Descartes viewed the existence of this gland in humans but not lesser animals as justification of his thesis that only humans have volition (Singer, 1928, p.128).

We can immediately see the possible challenge to realism; Iatrophysics postulates the existence of ‘Animal Spirits’, hollow nerve-tubes, and a motile pineal gland found exclusively in humans – all of which are rejected in modern physiology. If the programme is successful, it presents a possible counterexample to the realist thesis under the PRA, with Iatrophysics as T1, modern physiology as T2, and any of these rejected entities as e.

Before analysing whether Iatrophysics was successful, we must present the development of Descartes’ programme by post-Cartesian Iatrophysicists, notably including Giovanni Borelli, Jan Swammerdam and Niels Steensen (aka. Nicolaus Steno or Nicholas Stensen), amongst others. These empirical scientists were not committed to the grander Cartesian programme of using reason to deduce the nature of physiological phenomena (Grosholz, 1991, p.117), rather focussing upon quantitative experimentation (Shryock, 1961, p.221).

Jan Swammerdam developed the theory by analysing muscular contraction. He began from two premises implicit in Descartes’ work:

  1. All organs are mechanical, so replication of an input should cause an identical output (Cobb, 2002, p.399)
  2. There is no physical difference between voluntary and involuntary motion. (Swammerdam, 17586, p.125)

From these principles, he hypothesised that, contra Descartes (Cobb, 2002, p.397), muscle could be caused to contract artificially by stimulating its motor nerve. Furthermore, this could occur long after the organism’s death, and with muscles disconnected from the brain by amputation (Cobb, 2002, p.399). He called this phenomenon ‘irritability’ (Pubols, 1959, p.115). Swammerdam hypothesised this from premise (i), noting that artificially stimulating the nerve bypassed the brain and, if this was a purely corpuscular response, needed no psychological input (Swammerdam, 1758, p.124).

As muscles could be forced to contract repeatedly ex corpus, Swammerdam postulated that no extra bulk could be entering the muscle to cause contraction (ibid., p.126). Muscle contraction is therefore isovolumic, again contradicting Descartes’ inferences. Descartes believed that extra bulk from two sources accompanied contraction – Animal Spirits entering the muscle through the nerve-tube, and, in antagonistic pairs, matter transferred from the antagonist via a hypothesised “shunt” (Descartes, 1664, pp.24-8). Swammerdam rejected the former source of increased bulk due to repeated muscle contractility when no additional Animal Spirits could be entering the nerve, and the latter due to the ability of muscle to contract equally strongly, according to his theory of irritability, when disconnected from its antagonist (Swammerdam, 1758, p.124).

Borelli and Steensen developed theories of myological structure. Although they saw one another’s structures as incompatible rivals, Kardel shows that this was a mutual misunderstanding, and demonstrates that Steensen’s set of five core musculo-physical suppositions (Kardel, 1994, p.18) were consistent with Borelli’s simpler model (ibid., pp.33-48). Borelli (1680) presents his ideas as a geometric treatise, supported by mathematical lemmas and intricate physical calculations (see Des Chene, 2005, p.251). Borelli’s theory holds that muscle-flesh is composed of spongy fibres, which, contrary to then-established doctrine, run perpendicular to the direction of force (Des Chene, 2005, p.252). He gives mechanistic reasons for this supposition, demonstrating that any alternative alignment renders the muscle inadequate to exert great forces (Borelli, 1680, p.12). Using levers as a model, he hypothesised that a disproportionately large force is necessary to lift small weights (ibid., p.18).

Borelli also applied Galilean physics to animal bodies, innovating ‘centre-of-gravity’ calculations (Jaynes, 1970, p.232). Furthermore, he hypothesised the lung and thoracic volumes, and calculated the force of the heart necessary to alone produce Harvey’s double-circulatory system. As the required exertion of the heart significantly exceeds its capability, he hypothesised that the circular fibres of the arteries contract, smoothing and driving blood-flow – this is now established as the Windkessel Effect7 (Parker, 2009, p.112).

Steensen described a comprehensive myological theory, developing Borelli’s spongy fibres into a cross-woven mesh, offering stronger analysis of contractility (Scherz, 1971, pp.304-5). Both Borelli and Steensen’s theoretical models predict isovolumic muscle contraction (Kardel, 1994, p.19).

The constituent claims of, T1, Iatrophysics, include the Borelli-Steensen myology, Swammerdam’s muscle irritability thesis, ex corpus contractility, and isovolumic muscle contraction. T1 also includes Cartesian locomotion – the ‘heart-heat’ thesis, the existence of Animal Spirits, the hollow tube theory of nerve-structure, and the role of the pineal gland in voluntary motion. There is tension between the Cartesian and post-Cartesian aspects to T1 – Cartesian locomotion is at best tenuously compatible with isovolumic contraction and irritability. However, no post-Cartesian Iatrophysicist is known to have presented an alternative theory of nerve function, with Borelli, Steensen and Swammerdam, and their disciples, all assuming the Cartesian framework for their physiological theories. The programme also contains abortive attempts to generalise mechanical methodology to all organic functions. These were unsuccessful; the digestive and reproductive systems especially proved resistant to purely mechanistic analysis (Singer, 1928, p.129-31).

II.2: Iatrophysics: successfulness and maturity.

For this case-study to be relevant to HC, it is necessary to show Iatrophysics was mature and successful. Under the post-Cartesians, I claim, it was progressive and successful in several areas.

First, Swammerdam undertook impressive experimental confirmations of his theory of irritability and isovolumic contraction. Swammerdam conducted a paradigmatic and often-replicated experiment (see Holmes, 1993), the ‘Nerve-Muscle Preparation’. He amputated a frog thigh muscle and motor-nerve. When the nerve is stroked with scissors, the muscle contracts, demonstrating the predicted irritability. The effect is repeatable ad nauseum (Cobb, 2002, pp.397-400). This predictive success is particularly confirmatory for Iatrophysics as it demonstrates that the rival Animistic theory – that some ‘psyche’ force causes physiological phenomena (Hall, 1972a, pp.xxviii) – is untenable.

Swammerdam’s next experiment repeats the process, encased in an air-tight syringe with a water droplet placed at the opening (Kardel, 1994, p.16). When the muscle contracts, the droplet is unmoved, so muscle-volume has not increased, confirming isovolumic contraction. Again, the success is reinforced as it disconfirms a non-mechanistic hypothesis; the opposing Iatrochemical movement argued that contraction resulted from fermentation in the muscle, causing increased bulk. This hypothesis could not be reformulated consistently with the Swammerdam experiments (Jaynes, 1970, p.229).

Many replications were undertaken. Glisson demonstrated the result with a human arm in a water-bath (Jaynes, 1970, p.230), reinforcing the Iatrophysical hypothesis of physical equivalence between volitional and involuntary motion. Swammerdam reinforced the theory that the heart is merely a muscle by replicating the result with hearts8 (Cobb, 2002, p.398), and diaphragms (ibid., p.397). He used similar principles to demonstrate a coherent theory of penile erection by blood influx (Cobb, 2000, p.122). Furthermore, Swammerdam used this experimental data to hypothesise the later-identified distinct motor and sensory nerves (Pubols, 1959, p.114).

Borelli’s thesis of perpendicular fibrous muscular structure is confirmed by its predicted isovolumic contraction through the Swammerdam experiments. His contention that perpendicular fibres are necessary to explain the exertion of massive forces was subsequently corroborated by showing that the bicep exerts a 560lbs force to raise a 28lb weight in the hand (Des Chene, 2005, p.253). Borelli demonstrates these hypotheses by analysing flight, showing why birds can, and humans cannot, fly (Borelli, 1680, p.323). This insight, applied to the musculo-skeletal system, effectively states the Balance of Pauwels, 250-years before its full range of applications was realised (Maquet, 1992, p.335), and it was successfully used to treat fractures of the femoral neck (Maquet, 1980, p.237). Borelli also developed machines to measure and confirm his predictions of the lung and thoracic volumes, and blood-pressure, which confirmed his prediction of the Windkessel Effect (Parker, 2009, p.112).

The revolutionary mechanistic theory that the heart was purely a muscle was upheld and copiously confirmed by Steensen, who further demonstrated the power of his myology by successfully applying muscular principles to the diaphragm, oesophagus and tongue (Scherz, 1971, p.304).

Many constituents of T1 had use-novel predictive successes, which were subsequently empirically confirmed. Therefore, Iatrophysics should be classed as a early but mature and successful scientific research programme.

II.3: Undoing the Cartesian Bundle – identifying T1′

Descartes was undoubtedly committed to his programme en bloc (Brown, 1971, p.61). However, we can identify two distinct aspects of his Iatrophysics; following Des Chene (2005, p.246), I will call these ‘Mechanism as Ontology’ and ‘Mechanistic Method’. ‘Mechanism as Ontology’ states that an organism’s mechanistic properties exhaust the properties of the organism; organisms are purely machines. ‘Mechanistic method’ holds that the mechanical methodology of physical science can be fruitfully applicable to physiology to yield novel understandings of the functionality of physiological systems (ibid. p.249-50). The mechanistic method may seem minimal, even trivial, to modern eyes, but for Descartes and his contemporaries was controversial and underexplored. The ‘Cartesian Bundle’ is the unification of these two aspects into a scientific programme, with the restrictive ontology as presupposition, and mechanistic method as heuristic.

These aspects are separable. As Des Chene argues, Borelli9 and other post-Cartesians neve required and rarely adhered to the ontological restrictions of ‘Mechanism as Ontology’ (ibid. p.246). To justify accepting mechanistic method, all that is required is a weaker presupposition; that some organic processes are “sufficiently machine-like” (ibid. p.249) that mechanistic analysis produces truth-like insights. After ‘undoing the Cartesian bundle’, the post-Cartesians are committed only to this weaker claim in deriving their predictive successes.

To illustrate this point, consider a therapeutic application of Iatrophysics. Fever, according to Iatrophysicists Gideon Harvey and Andrew Brown, was caused by obstructions in the circulatory system – purely mechanical analysis yields all febrile symptoms in consequence (Sigal, 1978, pp.572-5). Cartesian ‘Mechanism as Ontology’ implies that such mechanical causes of fever are the only correct explanations for fever. The weaker claim, however, only implies that such obstructions can cause fever, but need not be the only cause. Modern fever theory concurs with this; Harvey and Brown discovered and treated embolic and thrombic fevers (Longmore, Wilkinson & Rajagopalan, 2004, p.552; Lippincott, Williams & Wilkins, 2008, pp.90-1), but there are many other types of fever not caused in this way (Fauci et al, 2009, pp.199-201) which would strictly speaking undermine the ‘Mechanism as Ontology’ thesis as understood by Descartes.

Hence, the programme’s predictive successes from the work of the post-Cartesians do not confirm the stronger ontological precepts. Meanwhile, the degenerative and highly unsuccessful attempts to generalise Iatrophysical principles to the digestive and reproductive systems, and Descartes’ fruitless attempts to develop an Iatrophysical theory of vision and emotion, derive from the ‘Mechanism as Ontology’ premise – but those failures are quite consistent with the mechanistic method. Hence, these ontological constituents of T1 are not part of T1′.

T1′ would incorporate the Borelli-Steensen myological theories, predicting isovolumic contraction, and Swammerdam’s irritability and ex corporeal contraction theses. The hypothesis that muscles are analogous to levers fuelled both the structural and force-exertion results. The thesis that the heart is a muscle was responsible for the prediction of the Windkessel effect and accurate calculation of blood pressure. Steensen’s development of a muscular analysis of the tongue, oesophagus and diaphragm were also successful. All of these constituent claims are elements of T1′.

None of the predictive successes offer any strong confirmation to the hypotheses that ‘Animal Spirits’ exist, the Pineal gland dictates voluntary action, or Cartesian nerve-structure. Animal Spirits and Cartesian neurology could only be made compatible with the experimental results of the post-Cartesians by ignoring their consequences as Borelli did (Des Chene, 2005, p.251), or supposing the nerve-transmission to be more like a vibration of the corpuscles, which approximates, as closely as pre-electrical science could, the modern action-potential theory, as Swammerdam did (Cobb, 2002, p.399). Hence, these postulates do not qualify for T1′. They were not elements of the derivation of the predictive successes of the theory, more inconvenient interpretative elements which the most successful Iatrophysicists learned to circumnavigate. The Pineal gland hypothesis was empirically disconfirmed when it was shown to be outside of the brain (Hall, 1972a, p.xxxvii) and not unique to humans (Jaynes, 1970, pp.226-7). The eventual excision of those postulates from medical science appears more as a natural consequence of the impediment they provided to the progress of the theory than a repudiation of core Iatrophysical theory.

II.4: Modern Compatibility – T1′ and T1*

The constituents identified in II.3 as T1′ are compatible with modern physiology, T2. First, the weak methodological claim; the flourishing biomechanics field illustrates that some organic phenomena approximate mechanical systems and are analysable using mechanical methodology (Maquet, 1992, p.339), so these are part of T2, and hence T1*. Modern physiology agrees that the heart, tongue and diaphragm are muscles, and that some muscles, like the biceps/triceps can be analysed by lever-approximations. The post-Cartesian ‘reflex-arc’ was integrated essentially unmodified into modern neuroscience by Hall in the 19th century (Singer, 1929, p.207-9). Irritability and ex corporeal contraction have both remained established physiological phenomena since Swammerdam’s demonstration, and his experiment and theory are taught in high school level biology (Holmes, 1993).

We turn, crucially, to Borelli and Steensen’s theories of muscular structure. Huijing (1995, p.451) shows that Borelli’s (1680) and Steensen’s (1667) theories are entirely accurate for some simple muscles, such shell molluscs. Lacking knowledge of the effects of elasticity is the main limitation in generalising their models. However, Huijing offers an extensive list of modern physiology texts employing Borelli and Steensen’s theories as approximations to modern myology. Kardel (1994) presents copious evidence of the compatibility of modern myology with Steensen’s theory. Pandy et al. (1990) used a musculo-skeletal theory derived from Borelli and Steensen, modified only to allow for tendon elasticity (Kardel, 1994, pp.46-51). So T1* incorporates the Iatrophysical myology.

In conclusion, the constituents of post-Cartesian Iatrophysics which were responsible for the programme’s successes are a subset of those compatible with modern physiology. Every claim identified as part of T1′ is an element of T1*. In this case study, HC holds.

III: A Critique of Psillos’ Realism

I argued that HC holds in the Iatrophysics case. HC is pivotal to Psillos’ position, but a lack of counterexamples alone cannot justify the conclusion that Psillos’ realism is adequate. I defend Psillos’ position against three objections: (i) the position is ad hoc, (ii) the divide et impera strategy is incompatible with realism, and (iii) we are unable to accurately decide the elements of T′.

(i): Ad-hocness

Psillos’ position is motivated to neutralise the PRA; therefore, his modification of SR3 into SR4 may appear ad hoc, circumventing the argument. Moreover, one may feel that the selectivity within Psillos’ realism is a disguise for a process of simply excluding counterexamples.

Against the former claim, I argue that Psillos’ reformulation of SR3 genuinely attempts to better characterise the realist’s intuitions. No realist should want to commit to the truthlikeness of every constituent of a theory because that theory was successful – there is a fallacy of composition and division embedded in the traditional realist position, overgeneralising the successes of T′ to become successes the theory T entire, and then redistributing the conferred truth-likeness across of every constituent of T. Psillos’ realism selects a subset T′ of T’s constituents which the realist believes are truth-like10.

If HC holds for a pair of theories T1 and T2 with respect to all entities e postulated by T1, the PRA cannot use that case to undermine SR4. The alleged ad hocness can then come in the selection of the subset T′. If T′ is deliberately constructed to exclude problematic constituents, then divide et impera cloaks an unwarranted evasion of the PRA. However, this argument fails. Avoiding the PRA clearly depends on HC, a strong and historically contingent hypothesis. The two subsets of T1 addressed in HC have independently testable criteria. Those constituents of T1 strongly confirmed by T1’s use-novel predictive successes form T1′. Those constituents of T1 which are compatible with T2 form T1*. The PRA, if it is to operate on SR4 at all, presupposes the existence of a pair of theories T1 and T2 such that some element(s) of T1′ are not in T1*. This is far from inconceivable – it requires detailed case-studies such as Section II to establish for each putative counterexample that these conditions are not fulfilled. Far from ruling out the PRA ad hoc, Psillos exposes SR4 to PRA, potentially, from any instance, and provides a bold criterion under which he would be forced to concede SR4.

Ad hoc strategies could absolutely be employed in any given realist defence of HC in a particular case. Defences of HC may use the benefit of hindsight to ascribe low significance to constituents of a theory which postulate entities later rejected, in an ahistorical way. While this would be poor scholarly practice, and rightly criticised in the specific instance, the openness to mistaken or ad hoc defences of HC does not itself render Psillos’ strategy ad hoc. We should be very cautious about accepting HC from swift or partial appraisals of theories, which lack engagement with or disregard the historical perspectives of the scientists responsible for the predictive successes of the theories under study.

(ii) ‘Divide et Impera’

Laudan (1981) anticipates a reformulation of SR3 towards SR4. He addresses Glymour’s attempts to “break out of this holist web” (Laudan, 1981, p.28) by assessing constituent claims individually. This move, and any like it – Divide et Impera for instance – are, Laudan argues, incompatible with realism. Laudan interprets Boyd (1973, p.1) and Sellars (1963, p.97) as maintaining that, for realism to operate, evidence for any part of a theory must be evidence for that theory in totality. Realists’ confirmation is all-or-nothing (Laudan, 1981, p.28). This, he argues, is the only way realists can support the inference to the truth-likeness of non-observational theoretical constituents of T by reference to T’s empirical successes. It is only with ‘Blockish Holism’, which Kitcher rejects, intact that realists can argue for truth-likeness beyond the lowest-level theoretical claims (ibid.). As Laudan extrapolates this argument from Boyd, I call its central thesis the Boyd-Laudan Conjecture (BLC):

BLC:   Evidence for T supports all constituents of T equally.

There is a two-pronged response to this attack; first, establishing that realism can operate without BLC, and second showing the falsity of BLC.

Stanford (2006) offers the first prong. Realism needs BLC if and only if there are no criteria to distinguish at every level the constituent claims which are confirmed by evidence E from those which are not (Stanford, 2006, pp.165-6). Psillos and Kitcher both offer putative criteria: Kitcher identifies the posits which do the “work” of predicting E as the ones confirmed by E, while Psillos similarly takes the constituents “essential” to the derivation of E as confirmed by E. Here, both are pursuing difficult lines. In (iii), I explore the difficulties in Psillos’ criteria. The lesson remains, though, that BLC is not necessary for realism, if some such attempt to provide rigorous criteria can succeed.

The second prong can be established by counterexamples to BLC. Consider “idle” posits. If a posit in T is idle, it can be altered without changing T’s empirical consequences. A notable example is the posit that the centre of the universe is at rest in Newtonian Mechanics. As Stanford argues (2006, pp.13-5), we can replace this claim with the claim that the centre of the universe moves at any given uniform velocity without any empirical consequences. There are infinitely many mutually exclusive equivalents to Newtonian mechanics of this form, all supported by identical evidence. Surely, then, the idle posit is not supported by the evidence for any of those Newtonian mechanics equivalents. The intuition is that Newtonian mechanics predicts all of the evidence in its favour without the idle claim, and so the claim is not supported by the evidence.

Worrall (1996) notes that idle posits in mature sciences are rare. Nevertheless, we can generalise this lesson through a modified Tacking Problem (see Chandler, 2007). Imagine a mature successful scientific theory T (Newtonian mechanics, say). Now create a new theory, {T ∪ X}, where X is a statement not in T, such that {T ∪ X} is empirically-equivalent to T. X could be a statement with no empirical consequences, like “Love will prevail” or the infamous ‘Nocturnal Doubling’ (Grunbaum, 1964)11 hypothesis. Given BLC, X now enjoys the same empirical confirmation as T by being conjoined to T in {T ∪ X}. But this is ridiculous – BLC cannot hold. BLC is neither true, nor necessary for realism. Therefore, the incompatibility of Divide et Impera and BLC is not a compelling challenge for Psillos.

(iii) Criteria for T

Psillos’ position relies upon the ability to complete his two divisions of T. However, our ability to determine the membership of T′ is questionable; Stanford (2003a; 2003b) argues that we historically fail to make the right division.

Stanford’s objection stems from Psillos’ observation, in support of his position, that working scientists express differentiated belief towards the constituent claims of their theories (Psillos, 1999, p.112). Stanford accepts this but argues that the constituents of T strongly believed by working scientists are often at variance to those the realist would, with hindsight, call T′. He cites passages from Maxwell, Weismann, and Lavoisier, showing endorsement of constituents of their theories which the modern realist might wish to exclude from T′ (Stanford, 2003b). Chang (2002) supports a similar interpretation of the attitudes of Lavoisier and Black in Psillos’ own case-study, the caloric theory of heat.

Furthermore, Psillos seems to expect an “implausible homogeneity” (Stanford, 2003a, p.923) in the attitudes of the various scientists within the field. Given that scientists’ attitudes will not be a reliable guide to which constituents of T should be deemed part of T′, Stanford envisions another historical induction12: from the past failure to correctly identify T′, we infer that our current decisions about which constituents of our theories are successful are also likely to be inadequate (Stanford, 2003b, p.569).

There are a number of issues with this objection. First, Psillos does not identify the working scientists as the group who should be executing the divide et impera. Rather, he references the actual scientific practice of differentiated belief to support the more primitive contention that our degrees-of-belief vary across individual constituent claims (Psillos, 1999, p.112).

Secondly, Stanford neglects the potential confounding factors in individual scientists’ attitudes – a scientist may apportion a degree-of-belief to a constituent claim which does not accurately reflect the confirmation afforded it by experimental evidence. Perhaps the scientist overlooks or overstates the case for a particular constituent because it contradicts or supports non-empirical beliefs – we see this in the case of Creation “Science” (cf. Kitcher, 1982). A rational self-interested scientist might overstate the case for a constituent which they formulated, or understate the case for a constituent challenging their own pet version of T.

The determination of T′ we seek is one purely based on the confirmation by T’s predictive successes. My case study supports this response: Descartes confidently believes anatomically untenable claims because they integrate with his metaphysics, a non-empirical motivation. Taking an individual scientist’s beliefs as the criteria for T′ – say, Lavoisier’s or Descartes’ – is unnecessary and unsustainable. Thus, we see that the disagreement between scientists highlighted by Stanford is actually a boon for Psillos; if we want to use scientists’ beliefs as an indicator of a claim’s confirmation, we hope to draw from a number of heterogeneous sources. Those claims strongly supported by disparate sources may warrant stronger belief.

Nevertheless, we still lack a systematic method for the decision of whether a claim t of T is in T′. Psillos presents a putative decision-procedure in terms of “essentiality” (Psillos, 1999, p.110), however this procedure quickly falls away from his discussion, preferring an intuitive appraisal in his own case studies (Psillos, 1994; 1999, pp.115-45). I followed this intuitive practice in Section II.3. Psillos wisely de-emphasises his essentiality criteria.13 Psillos seems justified, though, in relying upon an intuitive method of determining T′. Surely scientific theories are so varied that no single algorithm could be expected to accurately select the most successful parts of any given theory T. Saatsi argues that the realist should refuse to answer “singular questions” (Saatsi, 2009, p.362); we cannot always say whether we believe claim t to be approximately true or not – there are very complicated evidence/theory and intra-theoretic relationships to consider.14 Perhaps the sophisticated realist can begin to extend the differentiated belief model further – we may be more able to make specific claims of the form “p is a better approximation to the truth than q”. Even in the absence of guaranteed answers to ‘singular questions’, Psillos’ realism retains its support from the No Miracles Argument – the belief that some constituent claims of T are approximately true still explains the success of T, even if we cannot specify precisely the approximately-true subset of T.

IV: Conclusions

Psillos’ attempt to answer PMI and PRA relies upon the consistency of HC with historical fact. As one counterexample to HC undermines Psillos’ response to PRA, we must carefully consider potential counterexamples. The Iatrophysics case initially seems such an instance, as the seemingly central postulates of Animal Spirits, nerve-tubes and pineal motility have since been radically rejected. However, historical analysis shows that the theoretical commitments and constituent claims required by the scientists whose work fuelled the predictive successes of Iatrophysics exclude those components, which were unnecessary, and often impediments to, these successes. The predictive successes give the greatest confirmation to constituents of the Iatrophysics programme which are compatible with modern physiology.

Naturally we cannot thus conclude that HC holds universally. However, we do see that (i) constituent claims of a theory are not all supported equally by evidence, and (ii) when regarded through the lens of differentiated belief apportioned according to the confirmation of predictive successes, an apparently troublesome case actually supports Psillos’ position. There will be more tests for HC, but as yet the realist can occupy this intuitively plausible ground.

Psillos’ realism is imperfect; his criteria of essentiality, for instance, are superfluous and problematic. There is more to say in analysing differentiated belief and the responsibility model. Perhaps we should seek a fully graduated realism, under which the degree to which we are realist towards t, our confidence in the claim that “t is approximately true”, depends on the confirmatory evidence for t within the context of the predictive successes of a theory. This would likely offer a more sophisticated approach that Psillos’ selective realism, avoiding the arbitrariness that can result particularly at the borderlines of membership in T′. However, in the absence of such a model, Psillos’ realism remains coherently defensible despite PMI and PRA. Laudan’s argument (1981) alone cannot yet force the realist to surrender epistemic optimism.


  • Borelli, G.A. (1680) De Motu Animalium (Rome: Angelo Bernarbo), translation: Maquet, P. (trans.) (1989) ‘On the Movement of Animals’ (Heidelberg: Springer-Verlag)
  • Boyd, R. (1973) ‘Realism, Underdetermination and a Causal Theory of Evidence’, Nous, 7: 1-12
  • Brown, T.M. (1971) ‘Descartes, Rene Du Perron’, in Gillespie, C. (ed.) The Dictionary of Scientific Biography, (New York: Linda Hall Library), pp.51-73
  • Chandler, J. (2007) ‘Solving the Tacking Problem with Contrast Classes’, British Journal of Philosophy of Science, 58: 489-502
  • Chang, H. (2003) ‘Preservative Realism and Its Discontents: Revisiting Caloric’, Philosophy of Science, 70: (5) Proceedings of the 2002 Biennial Meeting of the Philosophy of Science Association. Part I: Contributed Papers pp.902-12
  • Cobb, M. (2002) ‘Exorcizing the Animal Spirits: Jan Swammerdam on nerve function’, Nature Reviews: Neuroscience, 3: 395-400
  • Descartes, R. (1664) Treatise of Man, trans: Hall,T.S. (1972) “French Text with Translation and Commentary by Thomas Steele Hall” (Cambridge, MA: Harvard UP)
  • Des Chene, D. (2005) ‘Mechanisms of life in the seventeenth century: Borelli, Perault, Regis’, Studies in History and Philosophy of Science; Part C: Studies in History and Philosophy of Biological and Biomedical Sciences, 36: (2) 245-260
  • Einstein, A. (1916) ‘The Foundation of the General Theory of Relativity’, in (1997) Kox, A.J., Klein, M.J. & Schulmann, R. (eds.) The Collected Papers of Albert Einstein: Volume 6: The Berlin Years: Writings, 1914-17, translation: Engel, A. (Princeton UP) pp.146-200.
  • Fauci, A.S. et al (2009) Harrison’s Manual of Medicine, 17th Edition (McGraw-Hill)
  • Grosholz, E.R. (1991) Cartesian Method and the Problem of Reduction (Oxford: Clarendon Press)
  • Grunbaum, A. (1964) ‘Is a universal nocturnal expansion falsifiable or physically vacuous?’, Philosophical Studies, 15: (5) 71-9
  • Hall, T.S. (1972a) “The Physiology of Descartes”, in Descartes, R. (1664) Treatise of Man, trans: Hall,T.S. (1972) “French Text with Translation and Commentary by Thomas Steele Hall” (Cambridge, MA: Harvard UP)
  • Hall, T.S. (1972b) “First French Edition: Synopsis of Contents”, in Descartes, R. (1664) Treatise of Man, trans: Hall,T.S. (1972) “French Text with Translation and Commentary by Thomas Steele Hall” (Harvard UP)
  • Hempel, C.G. (1970) ‘Studies in the Logic of Confirmation’, in Aspects of Scientific Explanation and Other Essays in the Philosophy of Science, (The Free Press) pp.3-52.
  • Holmes, F.L. (1993) ‘The old martyr of science: the frog in experimental physiology’, Journal of the History of Biology, 26: 311-28.
  • Jaynes, J. (1970) ‘The Problem of Animate Motion in the Seventeenth Century’, Journal of the History of Ideas, 31: (2) 219-34
  • Kardel, T. (1994) ‘Stensen’s myology in historical perspective’, in Transactions of the American Philosophical Society, New Series, 84: (1) ‘Steno on Muscles: Introduction, Texts, Translations’, pp. 1-vi
  • Kelly, B.A. & Chowienczyk, P. (2002) ‘Vascular Complience’, in Hunt, B.J., Poston, L., Schachter, M. & Halliday, A. (eds.) An Introduction to Vascular Biology (CUP)
  • Kitcher, P. (1982) Abusing Science (MIT Press)
  • Kitcher, P. (1993) The Advancement of Science (OUP)
  • Kitcher, P. (2001) ‘Real Realism: The Galilean Strategy’, Philosophical Review 110, April, pp151-197.
  • Lakatos, I. (1968) ‘Criticism and the Methodology of Scientific Research Programmes’, in Proceedings of the Aristotelian Society, New Series, 69: (1968-1969), pp. 149-186 (Blackwell)
  • Laudan, L. (1981) ‘A Confutation of Convergent Realism’, Philosophy of Science, 48, pp. 19-48
  • Lippincott, Williams & Wilkins (2008) Professional Guide to Diseases (Wolters Kluwer)
  • Longmore, M., Wilkinson, I.B. & Rajagopalan, S.R. (2004) Oxford Handbook of Clinical Medicine, Sixth Edition (OUP)
  • Lyons, T.D. (2006) ‘Scientific Realism and the Strategema de Divide et Impera’, British Journal of Philosophy of Science, 57: 537-60.
  • Lyons, T.D. (2009) ‘Non-competitor Conditions in the Scientific Realism Debate’, International Studies in the Philosophy of Science, 23: (1) 65-84
  • Maquet, P. (1980) ‘Friedrich Pauwels (1885-1980)’, International Orthopaedics, 4: (3) 237-8
  • Maquet, P. (1992) ‘Iatrophysics to Biomechanics: From Borelli (1608-1679) to Pauwels (1885-1980)’, The Journal of Joint and Bone Surgery 74-B: (3) 335-9
  • Pandy, M.G., Zajac, F.E. & Sim, E. et al (1990) ‘An Optimal-Control Model for Maximal-Height Human Jumping’, Journal of Biomechanics, 23: (12) 1185-98
  • Parker, K.H (2009) ‘A brief history of arterial wave mechanics’, Medical and Biological Engineering and Computing, 47: 111-8
  • Poincaré, H. (1952) Science and Hypothesis (New York: Dover), reprinted from Poincaré, H. (1902) La Science et l’hypothese.
  • Popper, K.R. (1972) ‘The Aim of Science’, in Objective Knowledge: an Evolutionary Approach (OUP) pp.191-205
  • Psillos, S. (1994) ‘A Philosophical Study of the Transition from the Caloric Theory of Heat to Thermodynamics: Resisting the Pessimistic Meta-Induction’, Studies in the History and Philosophy of Science, 25: 159-90
  • Psillos, S. (1996) ‘Scientific Realism and the ‘Pessimistic Induction’’, Philosophy of Science, 63: Supplement, Proceedings of the 1996 Biennial Meetings of the Philosophy of Science Association. Part I: Contributed Papers, pp.S306-14
  • Psillos, S. (1999) Scientific Realism: How Science Tracks Truth (Oxford: Routledge)
  • Psillos, S. (2009) Knowing the Structure of Nature: Essays on Realism and Explanation (McMillan)
  • Pubols, B. H. (1959) ‘Jan Swammerdam and the history of reflex action’, American Journal of Psychology, 72: 131-135.
  • Quin, C.E. (1997) ‘The ideas of Thomas Kuhn in relation to medical advances in the sixteenth and seventeenth centuries’, Journal of the Royal Society of Medicine, 90: 225-8
  • Saatsi, J. (2009) ‘Grasping at Realist Straws’, Metascience, 18: 355-62
  • Scherz, G. (1971) ‘Stenson, Nicholas’, in Gillespie, C. (ed.) The Dictionary of Scientific Biography, (New York: Linda Hall Library), pp.303-8
  • Sellars, W. (1963) Science, Perception and Reality (New York: Humanities Press)
  • Settle, T.B. (1971) ‘Borelli, Giovanni Alfonso’, in Gillespie, C. (ed.) The Dictionary of Scientific Biography, (New York: Linda Hall Library), pp.306-314
  • Shryock, R.H. (1961) ‘The History of Quantification in Medical Science’, Isis 52: (2) 215-37
  • Sigal, S.L. (1978) ‘Fever Theory in the Seventeenth Century: building toward a comprehensive physiology’, The Yale Journal of Biology and Medicine, 51: 571-82
  • Singer, C. (1928) A Short History of Medicine (Oxford: Clarendon Press)
  • Stanford, P.K. (2003a) ‘No Refuge for Realism: Selective Confirmation and the History of Science’, Philosophy of Science, 70: (5), Proceedings of the 2002 Biennial Meeting of the Philosophy of Science Association. Part I: Contributed Papers, pp. 913-925
  • Stanford, P.K. (2003b) ‘Pyrrhic Victories for Scientific Realism’, The Journal of Philosophy, 100: (11) 553-72
  • Stanford, P. K. (2006) Exceeding Our Grasp – Science, History, and the Problem of Unconceived Alternatives. (Oxford: OUP)
  • Starobinksi, J. (1964) A History of Medicine, trans: Swift, B.C. (London: Leisure Arts Ltd.)
  • Steensen, N. (1667) Elementorum Myologiae Specimen sev Musculi Descriptio Geometrica (Florence), translation: Kardel, T. (1994), Transactions of the American Philosophical Society, 84: 1-252
  • Swammerdam, J. (1758) The Book of Nature, or the History of Insects: Reduced to Distinct Classes. Confirmed by particular instances, Displayed in the Anatomical Analysis of many species. Flloyd, T. (trans.) & Hill, J. (ed.)
  • Van Fraassan, B.C. (1998) Arguments Concerning Scientific Realism, in Curd, M. & Cover, J.A. (eds) Philosophy of Science: The Central Issues, pp.1064-1087.
  • Windelspecht, M. (2002) Groundbreaking scientific experiments, inventions, and discoveries of the 17th century (Greenwood Publishing Group)
  • Worrall, J. (1994) ‘How to remain (reasonably) optimistic: Scientific Realism and the “luminiferous aether”’, PSA 1994, vol. 1, pp. 334-342
  • Worrall, J. (1996) ‘Thomas Young and the “Refutation” of Newtonian Optics’, in Howson, C. (ed) Method and Appraisal in the Physical Sciences (Cambridge, UK: CUP)
  • Worrall, J. (2002) ‘New Evidence For Old’, in Gardenfors, P., Wolenski, J. & Kijania-Placek, K. (eds.) In the Scope of Logic, Methodology and Philosophy of Science, 1, pp. 191-209. (Kluwer Academic Publishers)

This paper was adapted from an early paper by the author entitled ‘Scientific Realism and the History of Medicine’.

Featured image from Descartes (1664).

Latest revision: 13/10/2022

  1. The argument originates with Poincaré (1952).
  2. The notation HC, T1, T2, T1′ and T1* is my own to clarify the arguments, and does not appear in this form in Psillos.
  3. A red sphere confirms the hypothesis that all blue objects are cubes – this is a version of Hempel’s Raven argument (see Hempel, 1970, p.12). It is correct, assuming the bag contains only finitely-many objects, as every object observed which is not a non-blue cube narrows the range of possible counterexamples to (2) (ibid., p.18-9).
  4. T1′ is empty if T1 is not mature or successful. Hence, HC holds vacuously for all immature theories – thus, a sophisticated version of the maturity defence is an inherent consequence of Psillos’ position.
  5. Alternatively: If T is a mature successful scientific theory, then the claims in T′ are all at least approximately true.
  6. Swammerdam’s opus, The Book of Nature, was widely known but formally unpublished until 56 years after his death (Singer, 1928, p.122).
  7. For a modern presentation of the Windkessel effect, see Kelly & Chowienczyk, 2002, pp.34-5.
  8. Given this thesis, the fact that the heart can continue contracting ex-corpus fuelled the positive expectation of the result in Swammerdam’s first experiment.
  9. See Settle, 1971 and Windelspecht, 2002, p.97-8, for Borelli’s commitment to other methods than purely mechanical physiology.
  10. Note that, as Lyons (2009, p.67) argues, the sophisticated Scientific Realist believes not the elements of T′ themselves, rather the second-order thesis that the elements of T′ are approximately true.
  11. ‘Nocturnal Doubling’ is the hypothesis that everything in the universe instantaneously doubled in size at time t. Grunbaum (1964) shows the empirical-equivalence of T and {T ∪ X} where T is Newtonian mechanics and X is Nocturnal Doubling.
  12. Note that this novel historical induction performed by Stanford is not the same as his more widely discussed New Induction argument (for which see Stanford, 2006).
  13. See Lyons (2006, pp.539-43) for the flaws in Psillos’ definition of essentiality.
  14. For this reason, I prefer the formulation of SR4 given above to the alternative in n.4, which specifically refers to T′.