Categorial modal realism.

The current conception of the plurality of worlds is founded on a set theoretic understanding of possibilia. This paper provides an alternative category theoretic conception and argues that it is at least as serviceable for our understanding of possibilia. In addition to or instead of the notion of possibilia conceived as possible objects or possible individuals, this alternative to set theoretic modal realism requires the notion of possible morphisms, conceived as possible changes, processes or transformations. To support this alternative conception of the plurality of worlds, I provide two examples where a category theoretic account can do work traditionally done by the set theoretic account: one on modal logic and another on paradoxes of size. I argue that the categorial account works at least as well as the set theoretic account, and moreover suggest that it has something to add in each case: it makes apparent avenues of inquiry that were obscured, if not invisible, on the set theoretic account. I conclude with a plea for epistemological humility about our acceptance of either a category-like or set-like realist ontology of modality.

What is Interpretability?

We argue that artificial networks are explainable and offer a novel theory of interpretability. Two sets of conceptual questions are prominent in theoretical engagements with artificial neural networks, especially in the context of medical artificial intelligence: (1) Are networks explainable, and if so, what does it mean to explain the output of a network? And (2) what does it mean for a network to be interpretable? We argue that accounts of "explanation" tailored specifically to neural networks have ineffectively reinvented the wheel. In response to (1), we show how four familiar accounts of explanation apply to neural networks as they would to any scientific phenomenon. We diagnose the confusion about explaining neural networks within the machine learning literature as an equivocation on "explainability," "understandability" and "interpretability." To remedy this, we distinguish between these notions, and answer (2) by offering a theory and typology of interpretation in machine learning. Interpretation is something one does to an explanation with the aim of producing another, more understandable, explanation. As with explanation, there are various concepts and methods involved in interpretation: Total or Partial, Global or Local, and Approximative or Isomorphic. Our account of "interpretability" is consistent with uses in the machine learning literature, in keeping with the philosophy of explanation and understanding, and pays special attention to medical artificial intelligence systems.

The role of purifying selection in the origin and maintenance of complex function.

Fitness contribution alone should not be the criterion of 'function' in molecular biology and genomics. Disagreement over the use of 'function' in molecular biology and genomics is still with us, almost eight years after publicity surrounding the Encyclopedia of DNA Elements project claimed that 80.4% of the human genome comprises "functional elements". Recent approaches attempt to resolve or reformulate this debate by redefining genomic 'function' in terms of current fitness contribution. In its favour, this redefinition for the genomic context is in apparent conformity with predominant experimental practices, especially in biomedical research, and with ascription of function by selective maintenance. We argue against approaches of this kind, however, on the grounds that they could be seen as non-Darwinian, and fail to properly account for the diversity of non-adaptive processes involved in the origin and maintenance of genomic complexity. We examine cases of molecular and organismal complexity that arise neutrally, showing how purifying selection maintains non-adaptive genomic complexity. Rather than lumping different sorts of genomic complexity together by defining 'function' as fitness contribution, we argue that it is best to separate the heterogeneous contributions of preaptation, exaptation and adaptation to the historical processes of origin and maintenance for complex features.

On causal roles and selected effects: our genome is mostly junk.

The idea that much of our genome is irrelevant to fitness-is not the product of positive natural selection at the organismal level-remains viable. Claims to the contrary, and specifically that the notion of "junk DNA" should be abandoned, are based on conflating meanings of the word "function". Recent estimates suggest that perhaps 90% of our DNA, though biochemically active, does not contribute to fitness in any sequence-dependent way, and possibly in no way at all. Comparisons to vertebrates with much larger and smaller genomes (the lungfish and the pufferfish) strongly align with such a conclusion, as they have done for the last half-century.

What Is the Tree of Life?

A universal Tree of Life (TOL) has long been a goal of molecular phylogeneticists, but reticulation at the level of genes and possibly at the levels of cells and species renders any simple interpretation of such a TOL, especially as applied to prokaryotes, problematic.

Aims and methods of biosteganography.

Applications of biotechnology to information security are now possible and have potentially far reaching political and technological implications. This change in information security practices, initiated by advancements in molecular biological and biotechnology, warrants reasonable and widespread consideration by biologists, biotechnologists and philosophers. I offer an explication of the landmark contributions, developments and current possibilities of biosteganography-the process of transmitting secure messages via biological mediums. I address, (i) how information can be stored and encoded in biological mediums, (ii) how biological mediums (e.g. DNA, RNA, protein) and storage systems (e.g. cells, biofilms, organisms) influence the nature of information security, and (iii) what constitutes a viable application of such biotechnologies.

Multilevel Selection Theory and the Evolutionary Functions of Transposable Elements.

One of several issues at play in the renewed debate over "junk DNA" is the organizational level at which genomic features might be seen as selected, and thus to exhibit function, as etiologically defined. The intuition frequently expressed by molecular geneticists that junk DNA is functional because it serves to "speed evolution" or as an "evolutionary repository" could be recast as a claim about selection between species (or clades) rather than within them, but this is not often done. Here, we review general arguments for the importance of selection at levels above that of organisms in evolution, and develop them further for a common genomic feature: the carriage of transposable elements (TEs). In many species, not least our own, TEs comprise a large fraction of all nuclear DNA, and whether they individually or collectively contribute to fitness--or are instead junk--is a subject of ongoing contestation. Even if TEs generally owe their origin to selfish selection at the lowest level (that of genomes), their prevalence in extant organisms and the prevalence of extant organisms bearing them must also respond to selection within species (on organismal fitness) and between species (on rates of speciation and extinction). At an even higher level, the persistence of clades may be affected (positively or negatively) by TE carriage. If indeed TEs speed evolution, it is at these higher levels of selection that such a function might best be attributed to them as a class.

Distinguishing between "function" and "effect" in genome biology.

Much confusion in genome biology results from conflation of possible meanings of the word "function." We suggest that, in this connection, attention should be paid to evolutionary biologists and philosophers who have previously dealt with this problem. We need only decide that although all genomic structures have effects, only some of them should be said to have functions. Although it will very often be difficult or impossible to establish function (strictly defined), it should not automatically be assumed. We enjoin genomicists in particular to pay greater attention to parsing biological effects.